From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Tue, 10 Mar 2020 11:00:07 +0000 (+0000)
Subject: [CostModel][X86] Add fmaxnum/fminnum costs tests
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=0b1dc6016f151f1529e20d863162e128c1623ca7;p=platform%2Fupstream%2Fllvm.git

[CostModel][X86] Add fmaxnum/fminnum costs tests
---

diff --git a/llvm/test/Analysis/CostModel/X86/fmaxnum.ll b/llvm/test/Analysis/CostModel/X86/fmaxnum.ll
new file mode 100644
index 0000000..0b57780
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/fmaxnum.ll
@@ -0,0 +1,129 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE2
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+;
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=slm | FileCheck %s --check-prefixes=SLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s --check-prefixes=GLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,BTVER2
+
+define i32 @f32(i32 %arg) {
+; SSE2-LABEL: 'f32'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'f32'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<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.maxnum.v8f32(<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.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'f32'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'f32'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 196 for instruction: %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'f32'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<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.maxnum.v8f32(<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.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'f32'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<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.maxnum.v8f32(<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.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.maxnum.f32(float undef, float undef)
+  %V2F32 = call <2 x float> @llvm.maxnum.v2f32(<2 x float> undef, <2 x float> undef)
+  %V4F32 = call <4 x float> @llvm.maxnum.v4f32(<4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.maxnum.v8f32(<8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.maxnum.v16f32(<16 x float> undef, <16 x float> undef)
+  ret i32 undef
+}
+
+define i32 @f64(i32 %arg) {
+; SSE-LABEL: 'f64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4f64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8f64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'f64'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4f64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8f64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'f64'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4f64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8f64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'f64'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<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.maxnum.v4f64(<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.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'f64'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<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.maxnum.v4f64(<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.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+  %f64 = call double @llvm.maxnum.f64(double undef, double undef)
+  %V2f64 = call <2 x double> @llvm.maxnum.v2f64(<2 x double> undef, <2 x double> undef)
+  %V4f64 = call <4 x double> @llvm.maxnum.v4f64(<4 x double> undef, <4 x double> undef)
+  %V8f64 = call <8 x double> @llvm.maxnum.v8f64(<8 x double> undef, <8 x double> undef)
+  %V16f64 = call <16 x double> @llvm.maxnum.v16f64(<16 x double> undef, <16 x double> undef)
+  ret i32 undef
+}
+
+declare float @llvm.maxnum.f32(float, float)
+declare <2 x float> @llvm.maxnum.v2f32(<2 x float>, <2 x float>)
+declare <4 x float> @llvm.maxnum.v4f32(<4 x float>, <4 x float>)
+declare <8 x float> @llvm.maxnum.v8f32(<8 x float>, <8 x float>)
+declare <16 x float> @llvm.maxnum.v16f32(<16 x float>, <16 x float>)
+
+declare double @llvm.maxnum.f64(double, double)
+declare <2 x double> @llvm.maxnum.v2f64(<2 x double>, <2 x double>)
+declare <4 x double> @llvm.maxnum.v4f64(<4 x double>, <4 x double>)
+declare <8 x double> @llvm.maxnum.v8f64(<8 x double>, <8 x double>)
+declare <16 x double> @llvm.maxnum.v16f64(<16 x double>, <16 x double>)
diff --git a/llvm/test/Analysis/CostModel/X86/fminnum.ll b/llvm/test/Analysis/CostModel/X86/fminnum.ll
new file mode 100644
index 0000000..b677310
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/X86/fminnum.ll
@@ -0,0 +1,129 @@
+; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE2
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+;
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=slm | FileCheck %s --check-prefixes=SLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=goldmont | FileCheck %s --check-prefixes=GLM
+; RUN: opt < %s -cost-model -analyze -mtriple=x86_64-- -mcpu=btver2 | FileCheck %s --check-prefixes=CHECK,AVX,BTVER2
+
+define i32 @f32(i32 %arg) {
+; SSE2-LABEL: 'f32'
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SSE42-LABEL: 'f32'
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<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.minnum.v8f32(<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.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'f32'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'f32'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 196 for instruction: %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'f32'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<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.minnum.v8f32(<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.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'f32'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.minnum.f32(float undef, float undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.minnum.v4f32(<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.minnum.v8f32(<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.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+  %F32 = call float @llvm.minnum.f32(float undef, float undef)
+  %V2F32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
+  %V4F32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
+  %V8F32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
+  %V16F32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
+  ret i32 undef
+}
+
+define i32 @f64(i32 %arg) {
+; SSE-LABEL: 'f64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX-LABEL: 'f64'
+; AVX-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; AVX512-LABEL: 'f64'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; SLM-LABEL: 'f64'
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.minnum.v2f64(<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.minnum.v4f64(<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.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+; GLM-LABEL: 'f64'
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2f64 = call <2 x double> @llvm.minnum.v2f64(<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.minnum.v4f64(<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.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+;
+  %f64 = call double @llvm.minnum.f64(double undef, double undef)
+  %V2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
+  %V4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
+  %V8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
+  %V16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
+  ret i32 undef
+}
+
+declare float @llvm.minnum.f32(float, float)
+declare <2 x float> @llvm.minnum.v2f32(<2 x float>, <2 x float>)
+declare <4 x float> @llvm.minnum.v4f32(<4 x float>, <4 x float>)
+declare <8 x float> @llvm.minnum.v8f32(<8 x float>, <8 x float>)
+declare <16 x float> @llvm.minnum.v16f32(<16 x float>, <16 x float>)
+
+declare double @llvm.minnum.f64(double, double)
+declare <2 x double> @llvm.minnum.v2f64(<2 x double>, <2 x double>)
+declare <4 x double> @llvm.minnum.v4f64(<4 x double>, <4 x double>)
+declare <8 x double> @llvm.minnum.v8f64(<8 x double>, <8 x double>)
+declare <16 x double> @llvm.minnum.v16f64(<16 x double>, <16 x double>)