[CostModel][X86] Split int/fp arithmetic cost kinds tests

It's going to be much easier to maintain these tests (and all the check prefix combinations) if we don't mix cost kinds in the same file.

diff --git a/llvm/test/Analysis/CostModel/X86/arith-costkinds.ll b/llvm/test/Analysis/CostModel/X86/arith-costkinds.ll
deleted file mode 100644 (file)
index dfb0f89..0000000
--- a/llvm/test/Analysis/CostModel/X86/arith-costkinds.ll
+++ /dev/null
@@ -1,634 +0,0 @@
-; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=latency      < %s | FileCheck %s --check-prefixes=LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=SIZE_LATE
-
-define i32 @add(i32 %arg) {
-; LATE-LABEL: 'add'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'add'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'add'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = add i64 undef, undef
-  %V2I64 = add <2 x i64> undef, undef
-  %V4I64 = add <4 x i64> undef, undef
-  %V8I64 = add <8 x i64> undef, undef
-
-  %I32 = add i32 undef, undef
-  %V4I32 = add <4 x i32> undef, undef
-  %V8I32 = add <8 x i32> undef, undef
-  %V16I32 = add <16 x i32> undef, undef
-
-  %I16 = add i16 undef, undef
-  %V8I16 = add <8 x i16> undef, undef
-  %V16I16 = add <16 x i16> undef, undef
-  %V32I16 = add <32 x i16> undef, undef
-
-  %I8 = add i8 undef, undef
-  %V16I8 = add <16 x i8> undef, undef
-  %V32I8 = add <32 x i8> undef, undef
-  %V64I8 = add <64 x i8> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @sub(i32 %arg) {
-; LATE-LABEL: 'sub'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'sub'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'sub'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = sub i64 undef, undef
-  %V2I64 = sub <2 x i64> undef, undef
-  %V4I64 = sub <4 x i64> undef, undef
-  %V8I64 = sub <8 x i64> undef, undef
-
-  %I32 = sub i32 undef, undef
-  %V4I32 = sub <4 x i32> undef, undef
-  %V8I32 = sub <8 x i32> undef, undef
-  %V16I32 = sub <16 x i32> undef, undef
-
-  %I16 = sub i16 undef, undef
-  %V8I16 = sub <8 x i16> undef, undef
-  %V16I16 = sub <16 x i16> undef, undef
-  %V32I16 = sub <32 x i16> undef, undef
-
-  %I8 = sub i8 undef, undef
-  %V16I8 = sub <16 x i8> undef, undef
-  %V32I8 = sub <32 x i8> undef, undef
-  %V64I8 = sub <64 x i8> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @or(i32 %arg) {
-; LATE-LABEL: 'or'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'or'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'or'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = or i64 undef, undef
-  %V2I64 = or <2 x i64> undef, undef
-  %V4I64 = or <4 x i64> undef, undef
-  %V8I64 = or <8 x i64> undef, undef
-
-  %I32 = or i32 undef, undef
-  %V4I32 = or <4 x i32> undef, undef
-  %V8I32 = or <8 x i32> undef, undef
-  %V16I32 = or <16 x i32> undef, undef
-
-  %I16 = or i16 undef, undef
-  %V8I16 = or <8 x i16> undef, undef
-  %V16I16 = or <16 x i16> undef, undef
-  %V32I16 = or <32 x i16> undef, undef
-
-  %I8 = or i8 undef, undef
-  %V16I8 = or <16 x i8> undef, undef
-  %V32I8 = or <32 x i8> undef, undef
-  %V64I8 = or <64 x i8> undef, undef
-
-  %I1 = or i1 undef, undef
-  %V2I1 = or <2 x i1> undef, undef
-  %V4I1 = or <4 x i1> undef, undef
-  %V8I1 = or <8 x i1> undef, undef
-  %V16I1 = or <16 x i1> undef, undef
-  %V32I1 = or <32 x i1> undef, undef
-  %V64I1 = or <64 x i1> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @xor(i32 %arg) {
-; LATE-LABEL: 'xor'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'xor'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'xor'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = xor i64 undef, undef
-  %V2I64 = xor <2 x i64> undef, undef
-  %V4I64 = xor <4 x i64> undef, undef
-  %V8I64 = xor <8 x i64> undef, undef
-
-  %I32 = xor i32 undef, undef
-  %V4I32 = xor <4 x i32> undef, undef
-  %V8I32 = xor <8 x i32> undef, undef
-  %V16I32 = xor <16 x i32> undef, undef
-
-  %I16 = xor i16 undef, undef
-  %V8I16 = xor <8 x i16> undef, undef
-  %V16I16 = xor <16 x i16> undef, undef
-  %V32I16 = xor <32 x i16> undef, undef
-
-  %I8 = xor i8 undef, undef
-  %V16I8 = xor <16 x i8> undef, undef
-  %V32I8 = xor <32 x i8> undef, undef
-  %V64I8 = xor <64 x i8> undef, undef
-
-  %I1 = xor i1 undef, undef
-  %V2I1 = xor <2 x i1> undef, undef
-  %V4I1 = xor <4 x i1> undef, undef
-  %V8I1 = xor <8 x i1> undef, undef
-  %V16I1 = xor <16 x i1> undef, undef
-  %V32I1 = xor <32 x i1> undef, undef
-  %V64I1 = xor <64 x i1> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @and(i32 %arg) {
-; LATE-LABEL: 'and'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'and'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'and'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = and i64 undef, undef
-  %V2I64 = and <2 x i64> undef, undef
-  %V4I64 = and <4 x i64> undef, undef
-  %V8I64 = and <8 x i64> undef, undef
-
-  %I32 = and i32 undef, undef
-  %V4I32 = and <4 x i32> undef, undef
-  %V8I32 = and <8 x i32> undef, undef
-  %V16I32 = and <16 x i32> undef, undef
-
-  %I16 = and i16 undef, undef
-  %V8I16 = and <8 x i16> undef, undef
-  %V16I16 = and <16 x i16> undef, undef
-  %V32I16 = and <32 x i16> undef, undef
-
-  %I8 = and i8 undef, undef
-  %V16I8 = and <16 x i8> undef, undef
-  %V32I8 = and <32 x i8> undef, undef
-  %V64I8 = and <64 x i8> undef, undef
-
-  %I1 = and i1 undef, undef
-  %V2I1 = and <2 x i1> undef, undef
-  %V4I1 = and <4 x i1> undef, undef
-  %V8I1 = and <8 x i1> undef, undef
-  %V16I1 = and <16 x i1> undef, undef
-  %V32I1 = and <32 x i1> undef, undef
-  %V64I1 = and <64 x i1> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @mul(i32 %arg) {
-; LATE-LABEL: 'mul'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'mul'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'mul'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %I64 = mul i64 undef, undef
-  %V2I64 = mul <2 x i64> undef, undef
-  %V4I64 = mul <4 x i64> undef, undef
-  %V8I64 = mul <8 x i64> undef, undef
-
-  %I32 = mul i32 undef, undef
-  %V4I32 = mul <4 x i32> undef, undef
-  %V8I32 = mul <8 x i32> undef, undef
-  %V16I32 = mul <16 x i32> undef, undef
-
-  %I16 = mul i16 undef, undef
-  %V8I16 = mul <8 x i16> undef, undef
-  %V16I16 = mul <16 x i16> undef, undef
-  %V32I16 = mul <32 x i16> undef, undef
-
-  %I8 = mul i8 undef, undef
-  %V2I8 = mul <2 x i8> undef, undef
-  %V4I8 = mul <4 x i8> undef, undef
-  %V8I8 = mul <8 x i8> undef, undef
-  %V16I8 = mul <16 x i8> undef, undef
-  %V32I8 = mul <32 x i8> undef, undef
-  %V64I8 = mul <64 x i8> undef, undef
-
-  ret i32 undef
-}
-
-; A <2 x i32> gets expanded to a <2 x i64> vector.
-; A <2 x i64> vector multiply is implemented using
-; 3 PMULUDQ and 2 PADDS and 4 shifts.
-define void @mul_2i32() {
-; LATE-LABEL: 'mul_2i32'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
-;
-; SIZE-LABEL: 'mul_2i32'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
-;
-; SIZE_LATE-LABEL: 'mul_2i32'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
-;
-  %A0 = mul <2 x i32> undef, undef
-
-  ret void
-}

diff --git a/llvm/test/Analysis/CostModel/X86/arith-fp-codesize.ll b/llvm/test/Analysis/CostModel/X86/arith-fp-codesize.ll
new file mode 100644 (file)
index 0000000..5dccd4d
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-fp-codesize.ll
@@ -0,0 +1,602 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=-sse2 | FileCheck %s --check-prefixes=CHECK,SSE1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE42
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512
+;
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,GLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+
+define i32 @fadd(i32 %arg) {
+; CHECK-LABEL: 'fadd'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fadd <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fadd <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fadd float undef, undef
+  %V4F32 = fadd <4 x float> undef, undef
+  %V8F32 = fadd <8 x float> undef, undef
+  %V16F32 = fadd <16 x float> undef, undef
+
+  %F64 = fadd double undef, undef
+  %V2F64 = fadd <2 x double> undef, undef
+  %V4F64 = fadd <4 x double> undef, undef
+  %V8F64 = fadd <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsub(i32 %arg) {
+; CHECK-LABEL: 'fsub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float undef, undef
+  %V4F32 = fsub <4 x float> undef, undef
+  %V8F32 = fsub <8 x float> undef, undef
+  %V16F32 = fsub <16 x float> undef, undef
+
+  %F64 = fsub double undef, undef
+  %V2F64 = fsub <2 x double> undef, undef
+  %V4F64 = fsub <4 x double> undef, undef
+  %V8F64 = fsub <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg_idiom(i32 %arg) {
+; CHECK-LABEL: 'fneg_idiom'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float -0.0, undef
+  %V4F32 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V8F32 = fsub <8 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V16F32 = fsub <16 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+
+  %F64 = fsub double -0.0, undef
+  %V2F64 = fsub <2 x double> <double -0.0, double -0.0>, undef
+  %V4F64 = fsub <4 x double> <double -0.0, double -0.0, double -0.0, double -0.0>, undef
+  %V8F64 = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg(i32 %arg) {
+; CHECK-LABEL: 'fneg'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fneg float undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fneg <4 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fneg <8 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fneg <16 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fneg double undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fneg <2 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fneg <4 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fneg <8 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fneg float undef
+  %V4F32 = fneg <4 x float> undef
+  %V8F32 = fneg <8 x float> undef
+  %V16F32 = fneg <16 x float> undef
+
+  %F64 = fneg double undef
+  %V2F64 = fneg <2 x double> undef
+  %V4F64 = fneg <4 x double> undef
+  %V8F64 = fneg <8 x double> undef
+
+  ret i32 undef
+}
+
+define i32 @fmul(i32 %arg) {
+; CHECK-LABEL: 'fmul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fmul float undef, undef
+  %V4F32 = fmul <4 x float> undef, undef
+  %V8F32 = fmul <8 x float> undef, undef
+  %V16F32 = fmul <16 x float> undef, undef
+
+  %F64 = fmul double undef, undef
+  %V2F64 = fmul <2 x double> undef, undef
+  %V4F64 = fmul <4 x double> undef, undef
+  %V8F64 = fmul <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fdiv(i32 %arg) {
+; CHECK-LABEL: 'fdiv'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fdiv float undef, undef
+  %V4F32 = fdiv <4 x float> undef, undef
+  %V8F32 = fdiv <8 x float> undef, undef
+  %V16F32 = fdiv <16 x float> undef, undef
+
+  %F64 = fdiv double undef, undef
+  %V2F64 = fdiv <2 x double> undef, undef
+  %V4F64 = fdiv <4 x double> undef, undef
+  %V8F64 = fdiv <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @frem(i32 %arg) {
+; CHECK-LABEL: 'frem'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = frem float undef, undef
+  %V4F32 = frem <4 x float> undef, undef
+  %V8F32 = frem <8 x float> undef, undef
+  %V16F32 = frem <16 x float> undef, undef
+
+  %F64 = frem double undef, undef
+  %V2F64 = frem <2 x double> undef, undef
+  %V4F64 = frem <4 x double> undef, undef
+  %V8F64 = frem <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsqrt(i32 %arg) {
+; SSE1-LABEL: 'fsqrt'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fsqrt'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fsqrt'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX1-LABEL: 'fsqrt'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX2-LABEL: 'fsqrt'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fsqrt'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fsqrt'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fsqrt'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.sqrt.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.sqrt.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fabs(i32 %arg) {
+; SSE1-LABEL: 'fabs'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fabs'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fabs'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fabs'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fabs'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fabs'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fabs'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fabs.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.fabs.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fcopysign(i32 %arg) {
+; SSE1-LABEL: 'fcopysign'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fcopysign'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fcopysign'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fcopysign'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fcopysign'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fcopysign'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fcopysign'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.copysign.f32(float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.copysign.f64(double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fma(i32 %arg) {
+; SSE1-LABEL: 'fma'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fma'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fma'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fma'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fma'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fma'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fma'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+declare float @llvm.sqrt.f32(float)
+declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
+declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
+declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
+
+declare double @llvm.sqrt.f64(double)
+declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
+declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
+declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
+
+declare float @llvm.fabs.f32(float)
+declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
+declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
+declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
+
+declare double @llvm.fabs.f64(double)
+declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
+declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
+declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
+
+declare float @llvm.copysign.f32(float, float)
+declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
+declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
+declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
+
+declare double @llvm.copysign.f64(double, double)
+declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
+declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
+declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
+
+declare float @llvm.fma.f32(float, float, float)
+declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
+declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
+declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
+
+declare double @llvm.fma.f64(double, double, double)
+declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
+declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
+declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)

diff --git a/llvm/test/Analysis/CostModel/X86/arith-fp-costkinds.ll b/llvm/test/Analysis/CostModel/X86/arith-fp-costkinds.ll
deleted file mode 100644 (file)
index ce29e23..0000000
--- a/llvm/test/Analysis/CostModel/X86/arith-fp-costkinds.ll
+++ /dev/null
@@ -1,904 +0,0 @@
-; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,SSE-LATE,SSE2-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,SSE-SIZE,SSE2-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+sse2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,SSE-SIZE_LATE,SSE2-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX-LATE,AVX1-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX-SIZE,AVX1-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx  -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX-SIZE_LATE,AVX1-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX-LATE,AVX2-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX-SIZE,AVX2-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx2 -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX-SIZE_LATE,AVX2-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX512-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX512-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512f,+avx512vl -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX512-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,AVX512-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,AVX512-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mattr=+avx512bw,+avx512vl -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,AVX512-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,SSE-LATE,SLM-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,SSE-SIZE,SLM-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=slm -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,SSE-SIZE_LATE,SLM-SIZE_LATE
-;
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=latency      < %s | FileCheck %s --check-prefixes=CHECK,LATE,SSE-LATE,GLM-LATE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=code-size    < %s | FileCheck %s --check-prefixes=CHECK,SIZE,SSE-SIZE,GLM-SIZE
-; RUN: opt -mtriple=x86_64-- -passes="print<cost-model>" 2>&1 -disable-output -mcpu=goldmont -cost-kind=size-latency < %s | FileCheck %s --check-prefixes=CHECK,SIZE_LATE,SSE-SIZE_LATE,GLM-SIZE_LATE
-
-define i32 @fadd(i32 %arg) {
-; LATE-LABEL: 'fadd'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fadd float undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fadd <4 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fadd <8 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fadd <16 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fadd double undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fadd <2 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fadd <4 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fadd <8 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fadd'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fadd <16 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fadd <8 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fadd'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fadd <16 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fadd <8 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fadd float undef, undef
-  %V4F32 = fadd <4 x float> undef, undef
-  %V8F32 = fadd <8 x float> undef, undef
-  %V16F32 = fadd <16 x float> undef, undef
-
-  %F64 = fadd double undef, undef
-  %V2F64 = fadd <2 x double> undef, undef
-  %V4F64 = fadd <4 x double> undef, undef
-  %V8F64 = fadd <8 x double> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @fsub(i32 %arg) {
-; LATE-LABEL: 'fsub'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fsub float undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fsub <4 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fsub <8 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fsub <16 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fsub double undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fsub <2 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fsub <4 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fsub <8 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fsub'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fsub'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fsub float undef, undef
-  %V4F32 = fsub <4 x float> undef, undef
-  %V8F32 = fsub <8 x float> undef, undef
-  %V16F32 = fsub <16 x float> undef, undef
-
-  %F64 = fsub double undef, undef
-  %V2F64 = fsub <2 x double> undef, undef
-  %V4F64 = fsub <4 x double> undef, undef
-  %V8F64 = fsub <8 x double> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @fneg_idiom(i32 %arg) {
-; LATE-LABEL: 'fneg_idiom'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fsub float -0.000000e+00, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fsub double -0.000000e+00, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fneg_idiom'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fneg_idiom'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fsub float -0.0, undef
-  %V4F32 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, undef
-  %V8F32 = fsub <8 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
-  %V16F32 = fsub <16 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
-
-  %F64 = fsub double -0.0, undef
-  %V2F64 = fsub <2 x double> <double -0.0, double -0.0>, undef
-  %V4F64 = fsub <4 x double> <double -0.0, double -0.0, double -0.0, double -0.0>, undef
-  %V8F64 = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, undef
-
-  ret i32 undef
-}
-
-define i32 @fneg(i32 %arg) {
-; LATE-LABEL: 'fneg'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fneg float undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fneg <4 x float> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fneg <8 x float> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fneg <16 x float> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fneg double undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fneg <2 x double> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fneg <4 x double> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fneg <8 x double> undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fneg'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fneg float undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fneg <4 x float> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fneg <8 x float> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fneg <16 x float> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fneg double undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fneg <2 x double> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fneg <4 x double> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fneg <8 x double> undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fneg'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fneg float undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fneg <4 x float> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fneg <8 x float> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fneg <16 x float> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fneg double undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fneg <2 x double> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fneg <4 x double> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fneg <8 x double> undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fneg float undef
-  %V4F32 = fneg <4 x float> undef
-  %V8F32 = fneg <8 x float> undef
-  %V16F32 = fneg <16 x float> undef
-
-  %F64 = fneg double undef
-  %V2F64 = fneg <2 x double> undef
-  %V4F64 = fneg <4 x double> undef
-  %V8F64 = fneg <8 x double> undef
-
-  ret i32 undef
-}
-
-define i32 @fmul(i32 %arg) {
-; LATE-LABEL: 'fmul'
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fmul float undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fmul <4 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fmul <8 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fmul <16 x float> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fmul double undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fmul <2 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fmul <4 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fmul <8 x double> undef, undef
-; LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE-LABEL: 'fmul'
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
-; SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SIZE_LATE-LABEL: 'fmul'
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
-; SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fmul float undef, undef
-  %V4F32 = fmul <4 x float> undef, undef
-  %V8F32 = fmul <8 x float> undef, undef
-  %V16F32 = fmul <16 x float> undef, undef
-
-  %F64 = fmul double undef, undef
-  %V2F64 = fmul <2 x double> undef, undef
-  %V4F64 = fmul <4 x double> undef, undef
-  %V8F64 = fmul <8 x double> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @fdiv(i32 %arg) {
-; CHECK-LABEL: 'fdiv'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = fdiv float undef, undef
-  %V4F32 = fdiv <4 x float> undef, undef
-  %V8F32 = fdiv <8 x float> undef, undef
-  %V16F32 = fdiv <16 x float> undef, undef
-
-  %F64 = fdiv double undef, undef
-  %V2F64 = fdiv <2 x double> undef, undef
-  %V4F64 = fdiv <4 x double> undef, undef
-  %V8F64 = fdiv <8 x double> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @frem(i32 %arg) {
-; CHECK-LABEL: 'frem'
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = frem float undef, undef
-  %V4F32 = frem <4 x float> undef, undef
-  %V8F32 = frem <8 x float> undef, undef
-  %V16F32 = frem <16 x float> undef, undef
-
-  %F64 = frem double undef, undef
-  %V2F64 = frem <2 x double> undef, undef
-  %V4F64 = frem <4 x double> undef, undef
-  %V8F64 = frem <8 x double> undef, undef
-
-  ret i32 undef
-}
-
-define i32 @fsqrt(i32 %arg) {
-; SSE2-LATE-LABEL: 'fsqrt'
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SSE2-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE2-SIZE-LABEL: 'fsqrt'
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SSE2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE2-SIZE_LATE-LABEL: 'fsqrt'
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SSE2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX1-LATE-LABEL: 'fsqrt'
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX1-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX1-SIZE-LABEL: 'fsqrt'
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX1-SIZE_LATE-LABEL: 'fsqrt'
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX1-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX2-LATE-LABEL: 'fsqrt'
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX2-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX2-SIZE-LABEL: 'fsqrt'
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX2-SIZE_LATE-LABEL: 'fsqrt'
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX2-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-LATE-LABEL: 'fsqrt'
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE-LABEL: 'fsqrt'
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE_LATE-LABEL: 'fsqrt'
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SLM-LATE-LABEL: 'fsqrt'
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SLM-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SLM-SIZE-LABEL: 'fsqrt'
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SLM-SIZE_LATE-LABEL: 'fsqrt'
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; SLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; GLM-LATE-LABEL: 'fsqrt'
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; GLM-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; GLM-SIZE-LABEL: 'fsqrt'
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; GLM-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; GLM-SIZE_LATE-LABEL: 'fsqrt'
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-; GLM-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = call float @llvm.sqrt.f32(float undef)
-  %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
-  %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
-  %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
-
-  %F64 = call double @llvm.sqrt.f64(double undef)
-  %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
-  %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
-  %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
-
-  ret i32 undef
-}
-
-define i32 @fabs(i32 %arg) {
-; SSE-LATE-LABEL: 'fabs'
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE-LABEL: 'fabs'
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE_LATE-LABEL: 'fabs'
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-LATE-LABEL: 'fabs'
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE-LABEL: 'fabs'
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE_LATE-LABEL: 'fabs'
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-LATE-LABEL: 'fabs'
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE-LABEL: 'fabs'
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE_LATE-LABEL: 'fabs'
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = call float @llvm.fabs.f32(float undef)
-  %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
-  %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
-  %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
-
-  %F64 = call double @llvm.fabs.f64(double undef)
-  %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
-  %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
-  %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
-
-  ret i32 undef
-}
-
-define i32 @fcopysign(i32 %arg) {
-; SSE-LATE-LABEL: 'fcopysign'
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE-LABEL: 'fcopysign'
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE_LATE-LABEL: 'fcopysign'
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-LATE-LABEL: 'fcopysign'
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE-LABEL: 'fcopysign'
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE_LATE-LABEL: 'fcopysign'
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-LATE-LABEL: 'fcopysign'
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE-LABEL: 'fcopysign'
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE_LATE-LABEL: 'fcopysign'
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = call float @llvm.copysign.f32(float undef, float undef)
-  %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
-  %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
-  %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
-
-  %F64 = call double @llvm.copysign.f64(double undef, double undef)
-  %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
-  %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
-  %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
-
-  ret i32 undef
-}
-
-define i32 @fma(i32 %arg) {
-; SSE-LATE-LABEL: 'fma'
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; SSE-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE-LABEL: 'fma'
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; SSE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; SSE-SIZE_LATE-LABEL: 'fma'
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; SSE-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-LATE-LABEL: 'fma'
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE-LABEL: 'fma'
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX-SIZE_LATE-LABEL: 'fma'
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-LATE-LABEL: 'fma'
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX512-LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE-LABEL: 'fma'
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX512-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-; AVX512-SIZE_LATE-LABEL: 'fma'
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-; AVX512-SIZE_LATE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
-;
-  %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
-  %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
-  %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
-  %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
-
-  %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
-  %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
-  %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
-  %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
-
-  ret i32 undef
-}
-
-declare float @llvm.sqrt.f32(float)
-declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
-declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
-declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
-
-declare double @llvm.sqrt.f64(double)
-declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
-declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
-declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
-
-declare float @llvm.fabs.f32(float)
-declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
-declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
-declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
-
-declare double @llvm.fabs.f64(double)
-declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
-declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
-declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
-
-declare float @llvm.copysign.f32(float, float)
-declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
-declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
-declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
-
-declare double @llvm.copysign.f64(double, double)
-declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
-declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
-declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
-
-declare float @llvm.fma.f32(float, float, float)
-declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
-declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
-declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
-
-declare double @llvm.fma.f64(double, double, double)
-declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
-declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
-declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)

diff --git a/llvm/test/Analysis/CostModel/X86/arith-fp-latency.ll b/llvm/test/Analysis/CostModel/X86/arith-fp-latency.ll
new file mode 100644 (file)
index 0000000..4612836
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-fp-latency.ll
@@ -0,0 +1,602 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=-sse2 | FileCheck %s --check-prefixes=CHECK,SSE1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE42
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512
+;
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,GLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+
+define i32 @fadd(i32 %arg) {
+; CHECK-LABEL: 'fadd'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fadd float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fadd <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fadd <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fadd <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fadd double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fadd <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fadd <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fadd <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fadd float undef, undef
+  %V4F32 = fadd <4 x float> undef, undef
+  %V8F32 = fadd <8 x float> undef, undef
+  %V16F32 = fadd <16 x float> undef, undef
+
+  %F64 = fadd double undef, undef
+  %V2F64 = fadd <2 x double> undef, undef
+  %V4F64 = fadd <4 x double> undef, undef
+  %V8F64 = fadd <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsub(i32 %arg) {
+; CHECK-LABEL: 'fsub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fsub float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fsub <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fsub <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fsub <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fsub double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fsub <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fsub <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fsub <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float undef, undef
+  %V4F32 = fsub <4 x float> undef, undef
+  %V8F32 = fsub <8 x float> undef, undef
+  %V16F32 = fsub <16 x float> undef, undef
+
+  %F64 = fsub double undef, undef
+  %V2F64 = fsub <2 x double> undef, undef
+  %V4F64 = fsub <4 x double> undef, undef
+  %V8F64 = fsub <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg_idiom(i32 %arg) {
+; CHECK-LABEL: 'fneg_idiom'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fsub float -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fsub double -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float -0.0, undef
+  %V4F32 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V8F32 = fsub <8 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V16F32 = fsub <16 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+
+  %F64 = fsub double -0.0, undef
+  %V2F64 = fsub <2 x double> <double -0.0, double -0.0>, undef
+  %V4F64 = fsub <4 x double> <double -0.0, double -0.0, double -0.0, double -0.0>, undef
+  %V8F64 = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg(i32 %arg) {
+; CHECK-LABEL: 'fneg'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fneg float undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fneg <4 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fneg <8 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fneg <16 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fneg double undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fneg <2 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fneg <4 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fneg <8 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fneg float undef
+  %V4F32 = fneg <4 x float> undef
+  %V8F32 = fneg <8 x float> undef
+  %V16F32 = fneg <16 x float> undef
+
+  %F64 = fneg double undef
+  %V2F64 = fneg <2 x double> undef
+  %V4F64 = fneg <4 x double> undef
+  %V8F64 = fneg <8 x double> undef
+
+  ret i32 undef
+}
+
+define i32 @fmul(i32 %arg) {
+; CHECK-LABEL: 'fmul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F32 = fmul float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F32 = fmul <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = fmul <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16F32 = fmul <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %F64 = fmul double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F64 = fmul <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = fmul <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F64 = fmul <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fmul float undef, undef
+  %V4F32 = fmul <4 x float> undef, undef
+  %V8F32 = fmul <8 x float> undef, undef
+  %V16F32 = fmul <16 x float> undef, undef
+
+  %F64 = fmul double undef, undef
+  %V2F64 = fmul <2 x double> undef, undef
+  %V4F64 = fmul <4 x double> undef, undef
+  %V8F64 = fmul <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fdiv(i32 %arg) {
+; CHECK-LABEL: 'fdiv'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fdiv float undef, undef
+  %V4F32 = fdiv <4 x float> undef, undef
+  %V8F32 = fdiv <8 x float> undef, undef
+  %V16F32 = fdiv <16 x float> undef, undef
+
+  %F64 = fdiv double undef, undef
+  %V2F64 = fdiv <2 x double> undef, undef
+  %V4F64 = fdiv <4 x double> undef, undef
+  %V8F64 = fdiv <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @frem(i32 %arg) {
+; CHECK-LABEL: 'frem'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = frem float undef, undef
+  %V4F32 = frem <4 x float> undef, undef
+  %V8F32 = frem <8 x float> undef, undef
+  %V16F32 = frem <16 x float> undef, undef
+
+  %F64 = frem double undef, undef
+  %V2F64 = frem <2 x double> undef, undef
+  %V4F64 = frem <4 x double> undef, undef
+  %V8F64 = frem <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsqrt(i32 %arg) {
+; SSE1-LABEL: 'fsqrt'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fsqrt'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fsqrt'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX1-LABEL: 'fsqrt'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX2-LABEL: 'fsqrt'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fsqrt'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fsqrt'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fsqrt'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.sqrt.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.sqrt.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fabs(i32 %arg) {
+; SSE1-LABEL: 'fabs'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fabs'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fabs'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fabs'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fabs'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fabs'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fabs'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fabs.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.fabs.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fcopysign(i32 %arg) {
+; SSE1-LABEL: 'fcopysign'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fcopysign'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fcopysign'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fcopysign'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fcopysign'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fcopysign'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fcopysign'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.copysign.f32(float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.copysign.f64(double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fma(i32 %arg) {
+; SSE1-LABEL: 'fma'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fma'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fma'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fma'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fma'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fma'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fma'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+declare float @llvm.sqrt.f32(float)
+declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
+declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
+declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
+
+declare double @llvm.sqrt.f64(double)
+declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
+declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
+declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
+
+declare float @llvm.fabs.f32(float)
+declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
+declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
+declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
+
+declare double @llvm.fabs.f64(double)
+declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
+declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
+declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
+
+declare float @llvm.copysign.f32(float, float)
+declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
+declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
+declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
+
+declare double @llvm.copysign.f64(double, double)
+declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
+declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
+declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
+
+declare float @llvm.fma.f32(float, float, float)
+declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
+declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
+declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
+
+declare double @llvm.fma.f64(double, double, double)
+declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
+declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
+declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)

diff --git a/llvm/test/Analysis/CostModel/X86/arith-fp-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/arith-fp-sizelatency.ll
new file mode 100644 (file)
index 0000000..bbeff8a
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-fp-sizelatency.ll
@@ -0,0 +1,602 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=-sse2 | FileCheck %s --check-prefixes=CHECK,SSE1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE42
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512
+;
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=slm | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s --check-prefixes=CHECK,GLM
+; RUN: opt < %s -enable-no-nans-fp-math  -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+
+define i32 @fadd(i32 %arg) {
+; CHECK-LABEL: 'fadd'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fadd <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fadd <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fadd float undef, undef
+  %V4F32 = fadd <4 x float> undef, undef
+  %V8F32 = fadd <8 x float> undef, undef
+  %V16F32 = fadd <16 x float> undef, undef
+
+  %F64 = fadd double undef, undef
+  %V2F64 = fadd <2 x double> undef, undef
+  %V4F64 = fadd <4 x double> undef, undef
+  %V8F64 = fadd <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsub(i32 %arg) {
+; CHECK-LABEL: 'fsub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float undef, undef
+  %V4F32 = fsub <4 x float> undef, undef
+  %V8F32 = fsub <8 x float> undef, undef
+  %V16F32 = fsub <16 x float> undef, undef
+
+  %F64 = fsub double undef, undef
+  %V2F64 = fsub <2 x double> undef, undef
+  %V4F64 = fsub <4 x double> undef, undef
+  %V8F64 = fsub <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg_idiom(i32 %arg) {
+; CHECK-LABEL: 'fneg_idiom'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fsub float -0.0, undef
+  %V4F32 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V8F32 = fsub <8 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+  %V16F32 = fsub <16 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
+
+  %F64 = fsub double -0.0, undef
+  %V2F64 = fsub <2 x double> <double -0.0, double -0.0>, undef
+  %V4F64 = fsub <4 x double> <double -0.0, double -0.0, double -0.0, double -0.0>, undef
+  %V8F64 = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, undef
+
+  ret i32 undef
+}
+
+define i32 @fneg(i32 %arg) {
+; CHECK-LABEL: 'fneg'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fneg float undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fneg <4 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fneg <8 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fneg <16 x float> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fneg double undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fneg <2 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fneg <4 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fneg <8 x double> undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fneg float undef
+  %V4F32 = fneg <4 x float> undef
+  %V8F32 = fneg <8 x float> undef
+  %V16F32 = fneg <16 x float> undef
+
+  %F64 = fneg double undef
+  %V2F64 = fneg <2 x double> undef
+  %V4F64 = fneg <4 x double> undef
+  %V8F64 = fneg <8 x double> undef
+
+  ret i32 undef
+}
+
+define i32 @fmul(i32 %arg) {
+; CHECK-LABEL: 'fmul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fmul float undef, undef
+  %V4F32 = fmul <4 x float> undef, undef
+  %V8F32 = fmul <8 x float> undef, undef
+  %V16F32 = fmul <16 x float> undef, undef
+
+  %F64 = fmul double undef, undef
+  %V2F64 = fmul <2 x double> undef, undef
+  %V4F64 = fmul <4 x double> undef, undef
+  %V8F64 = fmul <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fdiv(i32 %arg) {
+; CHECK-LABEL: 'fdiv'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = fdiv float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fdiv <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fdiv <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fdiv <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = fdiv double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fdiv <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fdiv <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = fdiv float undef, undef
+  %V4F32 = fdiv <4 x float> undef, undef
+  %V8F32 = fdiv <8 x float> undef, undef
+  %V16F32 = fdiv <16 x float> undef, undef
+
+  %F64 = fdiv double undef, undef
+  %V2F64 = fdiv <2 x double> undef, undef
+  %V4F64 = fdiv <4 x double> undef, undef
+  %V8F64 = fdiv <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @frem(i32 %arg) {
+; CHECK-LABEL: 'frem'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F32 = frem float undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = frem <4 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = frem <8 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = frem <16 x float> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %F64 = frem double undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = frem <4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = frem <8 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = frem float undef, undef
+  %V4F32 = frem <4 x float> undef, undef
+  %V8F32 = frem <8 x float> undef, undef
+  %V16F32 = frem <16 x float> undef, undef
+
+  %F64 = frem double undef, undef
+  %V2F64 = frem <2 x double> undef, undef
+  %V4F64 = frem <4 x double> undef, undef
+  %V8F64 = frem <8 x double> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @fsqrt(i32 %arg) {
+; SSE1-LABEL: 'fsqrt'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fsqrt'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fsqrt'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX1-LABEL: 'fsqrt'
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX2-LABEL: 'fsqrt'
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fsqrt'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fsqrt'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fsqrt'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.sqrt.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.sqrt.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fabs(i32 %arg) {
+; SSE1-LABEL: 'fabs'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fabs'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fabs'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fabs'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fabs'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fabs'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fabs'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fabs.f32(float undef)
+  %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
+
+  %F64 = call double @llvm.fabs.f64(double undef)
+  %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fcopysign(i32 %arg) {
+; SSE1-LABEL: 'fcopysign'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fcopysign'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fcopysign'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fcopysign'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fcopysign'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fcopysign'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fcopysign'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.copysign.f32(float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.copysign.f64(double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+define i32 @fma(i32 %arg) {
+; SSE1-LABEL: 'fma'
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE2-LABEL: 'fma'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'fma'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'fma'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'fma'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'fma'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'fma'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
+  %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
+
+  %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
+  %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
+  %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
+  %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
+
+  ret i32 undef
+}
+
+declare float @llvm.sqrt.f32(float)
+declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
+declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
+declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
+
+declare double @llvm.sqrt.f64(double)
+declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
+declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
+declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
+
+declare float @llvm.fabs.f32(float)
+declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
+declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
+declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
+
+declare double @llvm.fabs.f64(double)
+declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
+declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
+declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
+
+declare float @llvm.copysign.f32(float, float)
+declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
+declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
+declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
+
+declare double @llvm.copysign.f64(double, double)
+declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
+declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
+declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
+
+declare float @llvm.fma.f32(float, float, float)
+declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
+declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
+declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
+
+declare double @llvm.fma.f64(double, double, double)
+declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
+declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
+declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)

diff --git a/llvm/test/Analysis/CostModel/X86/arith-int-codesize.ll b/llvm/test/Analysis/CostModel/X86/arith-int-codesize.ll
new file mode 100644 (file)
index 0000000..a8e506c
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-int-codesize.ll
@@ -0,0 +1,334 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+ssse3 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mattr=+avx512f,+avx512dq | FileCheck %s
+;
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=slm | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=code-size -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s
+
+define i32 @add(i32 %arg) {
+; CHECK-LABEL: 'add'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = add i64 undef, undef
+  %V2I64 = add <2 x i64> undef, undef
+  %V4I64 = add <4 x i64> undef, undef
+  %V8I64 = add <8 x i64> undef, undef
+
+  %I32 = add i32 undef, undef
+  %V4I32 = add <4 x i32> undef, undef
+  %V8I32 = add <8 x i32> undef, undef
+  %V16I32 = add <16 x i32> undef, undef
+
+  %I16 = add i16 undef, undef
+  %V8I16 = add <8 x i16> undef, undef
+  %V16I16 = add <16 x i16> undef, undef
+  %V32I16 = add <32 x i16> undef, undef
+
+  %I8 = add i8 undef, undef
+  %V16I8 = add <16 x i8> undef, undef
+  %V32I8 = add <32 x i8> undef, undef
+  %V64I8 = add <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @sub(i32 %arg) {
+; CHECK-LABEL: 'sub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = sub i64 undef, undef
+  %V2I64 = sub <2 x i64> undef, undef
+  %V4I64 = sub <4 x i64> undef, undef
+  %V8I64 = sub <8 x i64> undef, undef
+
+  %I32 = sub i32 undef, undef
+  %V4I32 = sub <4 x i32> undef, undef
+  %V8I32 = sub <8 x i32> undef, undef
+  %V16I32 = sub <16 x i32> undef, undef
+
+  %I16 = sub i16 undef, undef
+  %V8I16 = sub <8 x i16> undef, undef
+  %V16I16 = sub <16 x i16> undef, undef
+  %V32I16 = sub <32 x i16> undef, undef
+
+  %I8 = sub i8 undef, undef
+  %V16I8 = sub <16 x i8> undef, undef
+  %V32I8 = sub <32 x i8> undef, undef
+  %V64I8 = sub <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @or(i32 %arg) {
+; CHECK-LABEL: 'or'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = or i64 undef, undef
+  %V2I64 = or <2 x i64> undef, undef
+  %V4I64 = or <4 x i64> undef, undef
+  %V8I64 = or <8 x i64> undef, undef
+
+  %I32 = or i32 undef, undef
+  %V4I32 = or <4 x i32> undef, undef
+  %V8I32 = or <8 x i32> undef, undef
+  %V16I32 = or <16 x i32> undef, undef
+
+  %I16 = or i16 undef, undef
+  %V8I16 = or <8 x i16> undef, undef
+  %V16I16 = or <16 x i16> undef, undef
+  %V32I16 = or <32 x i16> undef, undef
+
+  %I8 = or i8 undef, undef
+  %V16I8 = or <16 x i8> undef, undef
+  %V32I8 = or <32 x i8> undef, undef
+  %V64I8 = or <64 x i8> undef, undef
+
+  %I1 = or i1 undef, undef
+  %V2I1 = or <2 x i1> undef, undef
+  %V4I1 = or <4 x i1> undef, undef
+  %V8I1 = or <8 x i1> undef, undef
+  %V16I1 = or <16 x i1> undef, undef
+  %V32I1 = or <32 x i1> undef, undef
+  %V64I1 = or <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @xor(i32 %arg) {
+; CHECK-LABEL: 'xor'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = xor i64 undef, undef
+  %V2I64 = xor <2 x i64> undef, undef
+  %V4I64 = xor <4 x i64> undef, undef
+  %V8I64 = xor <8 x i64> undef, undef
+
+  %I32 = xor i32 undef, undef
+  %V4I32 = xor <4 x i32> undef, undef
+  %V8I32 = xor <8 x i32> undef, undef
+  %V16I32 = xor <16 x i32> undef, undef
+
+  %I16 = xor i16 undef, undef
+  %V8I16 = xor <8 x i16> undef, undef
+  %V16I16 = xor <16 x i16> undef, undef
+  %V32I16 = xor <32 x i16> undef, undef
+
+  %I8 = xor i8 undef, undef
+  %V16I8 = xor <16 x i8> undef, undef
+  %V32I8 = xor <32 x i8> undef, undef
+  %V64I8 = xor <64 x i8> undef, undef
+
+  %I1 = xor i1 undef, undef
+  %V2I1 = xor <2 x i1> undef, undef
+  %V4I1 = xor <4 x i1> undef, undef
+  %V8I1 = xor <8 x i1> undef, undef
+  %V16I1 = xor <16 x i1> undef, undef
+  %V32I1 = xor <32 x i1> undef, undef
+  %V64I1 = xor <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @and(i32 %arg) {
+; CHECK-LABEL: 'and'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = and i64 undef, undef
+  %V2I64 = and <2 x i64> undef, undef
+  %V4I64 = and <4 x i64> undef, undef
+  %V8I64 = and <8 x i64> undef, undef
+
+  %I32 = and i32 undef, undef
+  %V4I32 = and <4 x i32> undef, undef
+  %V8I32 = and <8 x i32> undef, undef
+  %V16I32 = and <16 x i32> undef, undef
+
+  %I16 = and i16 undef, undef
+  %V8I16 = and <8 x i16> undef, undef
+  %V16I16 = and <16 x i16> undef, undef
+  %V32I16 = and <32 x i16> undef, undef
+
+  %I8 = and i8 undef, undef
+  %V16I8 = and <16 x i8> undef, undef
+  %V32I8 = and <32 x i8> undef, undef
+  %V64I8 = and <64 x i8> undef, undef
+
+  %I1 = and i1 undef, undef
+  %V2I1 = and <2 x i1> undef, undef
+  %V4I1 = and <4 x i1> undef, undef
+  %V8I1 = and <8 x i1> undef, undef
+  %V16I1 = and <16 x i1> undef, undef
+  %V32I1 = and <32 x i1> undef, undef
+  %V64I1 = and <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @mul(i32 %arg) {
+; CHECK-LABEL: 'mul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = mul i64 undef, undef
+  %V2I64 = mul <2 x i64> undef, undef
+  %V4I64 = mul <4 x i64> undef, undef
+  %V8I64 = mul <8 x i64> undef, undef
+
+  %I32 = mul i32 undef, undef
+  %V4I32 = mul <4 x i32> undef, undef
+  %V8I32 = mul <8 x i32> undef, undef
+  %V16I32 = mul <16 x i32> undef, undef
+
+  %I16 = mul i16 undef, undef
+  %V8I16 = mul <8 x i16> undef, undef
+  %V16I16 = mul <16 x i16> undef, undef
+  %V32I16 = mul <32 x i16> undef, undef
+
+  %I8 = mul i8 undef, undef
+  %V2I8 = mul <2 x i8> undef, undef
+  %V4I8 = mul <4 x i8> undef, undef
+  %V8I8 = mul <8 x i8> undef, undef
+  %V16I8 = mul <16 x i8> undef, undef
+  %V32I8 = mul <32 x i8> undef, undef
+  %V64I8 = mul <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+; A <2 x i32> gets expanded to a <2 x i64> vector.
+; A <2 x i64> vector multiply is implemented using
+; 3 PMULUDQ and 2 PADDS and 4 shifts.
+define void @mul_2i32() {
+; CHECK-LABEL: 'mul_2i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
+;
+  %A0 = mul <2 x i32> undef, undef
+
+  ret void
+}

diff --git a/llvm/test/Analysis/CostModel/X86/arith-int-latency.ll b/llvm/test/Analysis/CostModel/X86/arith-int-latency.ll
new file mode 100644 (file)
index 0000000..6d59a07
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-int-latency.ll
@@ -0,0 +1,334 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+ssse3 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mattr=+avx512f,+avx512dq | FileCheck %s
+;
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=slm | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=latency -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s
+
+define i32 @add(i32 %arg) {
+; CHECK-LABEL: 'add'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = add i64 undef, undef
+  %V2I64 = add <2 x i64> undef, undef
+  %V4I64 = add <4 x i64> undef, undef
+  %V8I64 = add <8 x i64> undef, undef
+
+  %I32 = add i32 undef, undef
+  %V4I32 = add <4 x i32> undef, undef
+  %V8I32 = add <8 x i32> undef, undef
+  %V16I32 = add <16 x i32> undef, undef
+
+  %I16 = add i16 undef, undef
+  %V8I16 = add <8 x i16> undef, undef
+  %V16I16 = add <16 x i16> undef, undef
+  %V32I16 = add <32 x i16> undef, undef
+
+  %I8 = add i8 undef, undef
+  %V16I8 = add <16 x i8> undef, undef
+  %V32I8 = add <32 x i8> undef, undef
+  %V64I8 = add <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @sub(i32 %arg) {
+; CHECK-LABEL: 'sub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = sub i64 undef, undef
+  %V2I64 = sub <2 x i64> undef, undef
+  %V4I64 = sub <4 x i64> undef, undef
+  %V8I64 = sub <8 x i64> undef, undef
+
+  %I32 = sub i32 undef, undef
+  %V4I32 = sub <4 x i32> undef, undef
+  %V8I32 = sub <8 x i32> undef, undef
+  %V16I32 = sub <16 x i32> undef, undef
+
+  %I16 = sub i16 undef, undef
+  %V8I16 = sub <8 x i16> undef, undef
+  %V16I16 = sub <16 x i16> undef, undef
+  %V32I16 = sub <32 x i16> undef, undef
+
+  %I8 = sub i8 undef, undef
+  %V16I8 = sub <16 x i8> undef, undef
+  %V32I8 = sub <32 x i8> undef, undef
+  %V64I8 = sub <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @or(i32 %arg) {
+; CHECK-LABEL: 'or'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = or i64 undef, undef
+  %V2I64 = or <2 x i64> undef, undef
+  %V4I64 = or <4 x i64> undef, undef
+  %V8I64 = or <8 x i64> undef, undef
+
+  %I32 = or i32 undef, undef
+  %V4I32 = or <4 x i32> undef, undef
+  %V8I32 = or <8 x i32> undef, undef
+  %V16I32 = or <16 x i32> undef, undef
+
+  %I16 = or i16 undef, undef
+  %V8I16 = or <8 x i16> undef, undef
+  %V16I16 = or <16 x i16> undef, undef
+  %V32I16 = or <32 x i16> undef, undef
+
+  %I8 = or i8 undef, undef
+  %V16I8 = or <16 x i8> undef, undef
+  %V32I8 = or <32 x i8> undef, undef
+  %V64I8 = or <64 x i8> undef, undef
+
+  %I1 = or i1 undef, undef
+  %V2I1 = or <2 x i1> undef, undef
+  %V4I1 = or <4 x i1> undef, undef
+  %V8I1 = or <8 x i1> undef, undef
+  %V16I1 = or <16 x i1> undef, undef
+  %V32I1 = or <32 x i1> undef, undef
+  %V64I1 = or <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @xor(i32 %arg) {
+; CHECK-LABEL: 'xor'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = xor i64 undef, undef
+  %V2I64 = xor <2 x i64> undef, undef
+  %V4I64 = xor <4 x i64> undef, undef
+  %V8I64 = xor <8 x i64> undef, undef
+
+  %I32 = xor i32 undef, undef
+  %V4I32 = xor <4 x i32> undef, undef
+  %V8I32 = xor <8 x i32> undef, undef
+  %V16I32 = xor <16 x i32> undef, undef
+
+  %I16 = xor i16 undef, undef
+  %V8I16 = xor <8 x i16> undef, undef
+  %V16I16 = xor <16 x i16> undef, undef
+  %V32I16 = xor <32 x i16> undef, undef
+
+  %I8 = xor i8 undef, undef
+  %V16I8 = xor <16 x i8> undef, undef
+  %V32I8 = xor <32 x i8> undef, undef
+  %V64I8 = xor <64 x i8> undef, undef
+
+  %I1 = xor i1 undef, undef
+  %V2I1 = xor <2 x i1> undef, undef
+  %V4I1 = xor <4 x i1> undef, undef
+  %V8I1 = xor <8 x i1> undef, undef
+  %V16I1 = xor <16 x i1> undef, undef
+  %V32I1 = xor <32 x i1> undef, undef
+  %V64I1 = xor <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @and(i32 %arg) {
+; CHECK-LABEL: 'and'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = and i64 undef, undef
+  %V2I64 = and <2 x i64> undef, undef
+  %V4I64 = and <4 x i64> undef, undef
+  %V8I64 = and <8 x i64> undef, undef
+
+  %I32 = and i32 undef, undef
+  %V4I32 = and <4 x i32> undef, undef
+  %V8I32 = and <8 x i32> undef, undef
+  %V16I32 = and <16 x i32> undef, undef
+
+  %I16 = and i16 undef, undef
+  %V8I16 = and <8 x i16> undef, undef
+  %V16I16 = and <16 x i16> undef, undef
+  %V32I16 = and <32 x i16> undef, undef
+
+  %I8 = and i8 undef, undef
+  %V16I8 = and <16 x i8> undef, undef
+  %V32I8 = and <32 x i8> undef, undef
+  %V64I8 = and <64 x i8> undef, undef
+
+  %I1 = and i1 undef, undef
+  %V2I1 = and <2 x i1> undef, undef
+  %V4I1 = and <4 x i1> undef, undef
+  %V8I1 = and <8 x i1> undef, undef
+  %V16I1 = and <16 x i1> undef, undef
+  %V32I1 = and <32 x i1> undef, undef
+  %V64I1 = and <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @mul(i32 %arg) {
+; CHECK-LABEL: 'mul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = mul i64 undef, undef
+  %V2I64 = mul <2 x i64> undef, undef
+  %V4I64 = mul <4 x i64> undef, undef
+  %V8I64 = mul <8 x i64> undef, undef
+
+  %I32 = mul i32 undef, undef
+  %V4I32 = mul <4 x i32> undef, undef
+  %V8I32 = mul <8 x i32> undef, undef
+  %V16I32 = mul <16 x i32> undef, undef
+
+  %I16 = mul i16 undef, undef
+  %V8I16 = mul <8 x i16> undef, undef
+  %V16I16 = mul <16 x i16> undef, undef
+  %V32I16 = mul <32 x i16> undef, undef
+
+  %I8 = mul i8 undef, undef
+  %V2I8 = mul <2 x i8> undef, undef
+  %V4I8 = mul <4 x i8> undef, undef
+  %V8I8 = mul <8 x i8> undef, undef
+  %V16I8 = mul <16 x i8> undef, undef
+  %V32I8 = mul <32 x i8> undef, undef
+  %V64I8 = mul <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+; A <2 x i32> gets expanded to a <2 x i64> vector.
+; A <2 x i64> vector multiply is implemented using
+; 3 PMULUDQ and 2 PADDS and 4 shifts.
+define void @mul_2i32() {
+; CHECK-LABEL: 'mul_2i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
+;
+  %A0 = mul <2 x i32> undef, undef
+
+  ret void
+}

diff --git a/llvm/test/Analysis/CostModel/X86/arith-int-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/arith-int-sizelatency.ll
new file mode 100644 (file)
index 0000000..e963fa0
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/arith-int-sizelatency.ll
@@ -0,0 +1,334 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+ssse3 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mattr=+avx512f,+avx512dq | FileCheck %s
+;
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=slm | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -cost-kind=size-latency -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s
+
+define i32 @add(i32 %arg) {
+; CHECK-LABEL: 'add'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = add i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = add <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = add <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = add <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = add i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = add <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = add <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = add <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = add i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = add <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = add <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = add <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = add i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = add <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = add <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = add <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = add i64 undef, undef
+  %V2I64 = add <2 x i64> undef, undef
+  %V4I64 = add <4 x i64> undef, undef
+  %V8I64 = add <8 x i64> undef, undef
+
+  %I32 = add i32 undef, undef
+  %V4I32 = add <4 x i32> undef, undef
+  %V8I32 = add <8 x i32> undef, undef
+  %V16I32 = add <16 x i32> undef, undef
+
+  %I16 = add i16 undef, undef
+  %V8I16 = add <8 x i16> undef, undef
+  %V16I16 = add <16 x i16> undef, undef
+  %V32I16 = add <32 x i16> undef, undef
+
+  %I8 = add i8 undef, undef
+  %V16I8 = add <16 x i8> undef, undef
+  %V32I8 = add <32 x i8> undef, undef
+  %V64I8 = add <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @sub(i32 %arg) {
+; CHECK-LABEL: 'sub'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = sub i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = sub <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = sub <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = sub <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = sub i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = sub <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = sub <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = sub <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = sub i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = sub <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = sub <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = sub <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = sub i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = sub <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = sub <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = sub <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = sub i64 undef, undef
+  %V2I64 = sub <2 x i64> undef, undef
+  %V4I64 = sub <4 x i64> undef, undef
+  %V8I64 = sub <8 x i64> undef, undef
+
+  %I32 = sub i32 undef, undef
+  %V4I32 = sub <4 x i32> undef, undef
+  %V8I32 = sub <8 x i32> undef, undef
+  %V16I32 = sub <16 x i32> undef, undef
+
+  %I16 = sub i16 undef, undef
+  %V8I16 = sub <8 x i16> undef, undef
+  %V16I16 = sub <16 x i16> undef, undef
+  %V32I16 = sub <32 x i16> undef, undef
+
+  %I8 = sub i8 undef, undef
+  %V16I8 = sub <16 x i8> undef, undef
+  %V32I8 = sub <32 x i8> undef, undef
+  %V64I8 = sub <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @or(i32 %arg) {
+; CHECK-LABEL: 'or'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = or i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = or <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = or <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = or <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = or i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = or <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = or <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = or <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = or i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = or <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = or <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = or <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = or i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = or <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = or <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = or <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = or i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = or <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = or <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = or <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = or <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = or <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = or <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = or i64 undef, undef
+  %V2I64 = or <2 x i64> undef, undef
+  %V4I64 = or <4 x i64> undef, undef
+  %V8I64 = or <8 x i64> undef, undef
+
+  %I32 = or i32 undef, undef
+  %V4I32 = or <4 x i32> undef, undef
+  %V8I32 = or <8 x i32> undef, undef
+  %V16I32 = or <16 x i32> undef, undef
+
+  %I16 = or i16 undef, undef
+  %V8I16 = or <8 x i16> undef, undef
+  %V16I16 = or <16 x i16> undef, undef
+  %V32I16 = or <32 x i16> undef, undef
+
+  %I8 = or i8 undef, undef
+  %V16I8 = or <16 x i8> undef, undef
+  %V32I8 = or <32 x i8> undef, undef
+  %V64I8 = or <64 x i8> undef, undef
+
+  %I1 = or i1 undef, undef
+  %V2I1 = or <2 x i1> undef, undef
+  %V4I1 = or <4 x i1> undef, undef
+  %V8I1 = or <8 x i1> undef, undef
+  %V16I1 = or <16 x i1> undef, undef
+  %V32I1 = or <32 x i1> undef, undef
+  %V64I1 = or <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @xor(i32 %arg) {
+; CHECK-LABEL: 'xor'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = xor i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = xor <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = xor <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = xor <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = xor i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = xor <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = xor <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = xor <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = xor i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = xor <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = xor <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = xor <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = xor i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = xor <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = xor <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = xor <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = xor i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = xor <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = xor <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = xor <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = xor <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = xor <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = xor <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = xor i64 undef, undef
+  %V2I64 = xor <2 x i64> undef, undef
+  %V4I64 = xor <4 x i64> undef, undef
+  %V8I64 = xor <8 x i64> undef, undef
+
+  %I32 = xor i32 undef, undef
+  %V4I32 = xor <4 x i32> undef, undef
+  %V8I32 = xor <8 x i32> undef, undef
+  %V16I32 = xor <16 x i32> undef, undef
+
+  %I16 = xor i16 undef, undef
+  %V8I16 = xor <8 x i16> undef, undef
+  %V16I16 = xor <16 x i16> undef, undef
+  %V32I16 = xor <32 x i16> undef, undef
+
+  %I8 = xor i8 undef, undef
+  %V16I8 = xor <16 x i8> undef, undef
+  %V32I8 = xor <32 x i8> undef, undef
+  %V64I8 = xor <64 x i8> undef, undef
+
+  %I1 = xor i1 undef, undef
+  %V2I1 = xor <2 x i1> undef, undef
+  %V4I1 = xor <4 x i1> undef, undef
+  %V8I1 = xor <8 x i1> undef, undef
+  %V16I1 = xor <16 x i1> undef, undef
+  %V32I1 = xor <32 x i1> undef, undef
+  %V64I1 = xor <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @and(i32 %arg) {
+; CHECK-LABEL: 'and'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = and i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = and <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = and <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = and <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = and i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = and <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = and <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = and <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = and i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = and <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = and <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = and <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = and i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = and <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = and <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = and <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I1 = and i1 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I1 = and <2 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I1 = and <4 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I1 = and <8 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I1 = and <16 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I1 = and <32 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I1 = and <64 x i1> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = and i64 undef, undef
+  %V2I64 = and <2 x i64> undef, undef
+  %V4I64 = and <4 x i64> undef, undef
+  %V8I64 = and <8 x i64> undef, undef
+
+  %I32 = and i32 undef, undef
+  %V4I32 = and <4 x i32> undef, undef
+  %V8I32 = and <8 x i32> undef, undef
+  %V16I32 = and <16 x i32> undef, undef
+
+  %I16 = and i16 undef, undef
+  %V8I16 = and <8 x i16> undef, undef
+  %V16I16 = and <16 x i16> undef, undef
+  %V32I16 = and <32 x i16> undef, undef
+
+  %I8 = and i8 undef, undef
+  %V16I8 = and <16 x i8> undef, undef
+  %V32I8 = and <32 x i8> undef, undef
+  %V64I8 = and <64 x i8> undef, undef
+
+  %I1 = and i1 undef, undef
+  %V2I1 = and <2 x i1> undef, undef
+  %V4I1 = and <4 x i1> undef, undef
+  %V8I1 = and <8 x i1> undef, undef
+  %V16I1 = and <16 x i1> undef, undef
+  %V32I1 = and <32 x i1> undef, undef
+  %V64I1 = and <64 x i1> undef, undef
+
+  ret i32 undef
+}
+
+define i32 @mul(i32 %arg) {
+; CHECK-LABEL: 'mul'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = mul <16 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = mul i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = mul <8 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = mul <16 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = mul <32 x i16> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = mul i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I8 = mul <2 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I8 = mul <4 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I8 = mul <8 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I8 = mul <16 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = mul <32 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64I8 = mul <64 x i8> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
+;
+  %I64 = mul i64 undef, undef
+  %V2I64 = mul <2 x i64> undef, undef
+  %V4I64 = mul <4 x i64> undef, undef
+  %V8I64 = mul <8 x i64> undef, undef
+
+  %I32 = mul i32 undef, undef
+  %V4I32 = mul <4 x i32> undef, undef
+  %V8I32 = mul <8 x i32> undef, undef
+  %V16I32 = mul <16 x i32> undef, undef
+
+  %I16 = mul i16 undef, undef
+  %V8I16 = mul <8 x i16> undef, undef
+  %V16I16 = mul <16 x i16> undef, undef
+  %V32I16 = mul <32 x i16> undef, undef
+
+  %I8 = mul i8 undef, undef
+  %V2I8 = mul <2 x i8> undef, undef
+  %V4I8 = mul <4 x i8> undef, undef
+  %V8I8 = mul <8 x i8> undef, undef
+  %V16I8 = mul <16 x i8> undef, undef
+  %V32I8 = mul <32 x i8> undef, undef
+  %V64I8 = mul <64 x i8> undef, undef
+
+  ret i32 undef
+}
+
+; A <2 x i32> gets expanded to a <2 x i64> vector.
+; A <2 x i64> vector multiply is implemented using
+; 3 PMULUDQ and 2 PADDS and 4 shifts.
+define void @mul_2i32() {
+; CHECK-LABEL: 'mul_2i32'
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %A0 = mul <2 x i32> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
+;
+  %A0 = mul <2 x i32> undef, undef
+
+  ret void
+}

diff --git a/llvm/test/Analysis/CostModel/X86/arith.ll b/llvm/test/Analysis/CostModel/X86/arith-int.ll
similarity index 100%
rename from llvm/test/Analysis/CostModel/X86/arith.ll
rename to llvm/test/Analysis/CostModel/X86/arith-int.ll