[X86][SSE] Remove use of SSE ADDS/SUBS saturation intrinsics from schedule/stack...

These are due to be upgraded soon, but good to replace them with generic llvm sadd_sat/ssub_sat intrinsics now.

The avx512 masked cases need doing as well but require a bit of tidyup first.

llvm-svn: 349621

diff --git a/llvm/test/CodeGen/X86/avx2-schedule.ll b/llvm/test/CodeGen/X86/avx2-schedule.ll
index e76ec04..00c4fcb 100644 (file)
--- a/llvm/test/CodeGen/X86/avx2-schedule.ll
+++ b/llvm/test/CodeGen/X86/avx2-schedule.ll
@@ -1129,12 +1129,12 @@ define <32 x i8> @test_paddsb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
 ; ZNVER1-NEXT:    vpaddsb %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpaddsb (%rdi), %ymm0, %ymm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8> %a0, <32 x i8> %a1)
+  %1 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   %2 = load <32 x i8>, <32 x i8> *%a2, align 32
-  %3 = call <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8> %1, <32 x i8> %2)
+  %3 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %1, <32 x i8> %2)
   ret <32 x i8> %3
 }
-declare <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <16 x i16> @test_paddsw(<16 x i16> %a0, <16 x i16> %a1, <16 x i16> *%a2) {
 ; GENERIC-LABEL: test_paddsw:
@@ -1172,12 +1172,12 @@ define <16 x i16> @test_paddsw(<16 x i16> %a0, <16 x i16> %a1, <16 x i16> *%a2)
 ; ZNVER1-NEXT:    vpaddsw %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpaddsw (%rdi), %ymm0, %ymm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16> %a0, <16 x i16> %a1)
+  %1 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   %2 = load <16 x i16>, <16 x i16> *%a2, align 32
-  %3 = call <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16> %1, <16 x i16> %2)
+  %3 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %1, <16 x i16> %2)
   ret <16 x i16> %3
 }
-declare <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <32 x i8> @test_paddusb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
 ; GENERIC-LABEL: test_paddusb:
@@ -6505,12 +6505,12 @@ define <32 x i8> @test_psubsb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
 ; ZNVER1-NEXT:    vpsubsb %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpsubsb (%rdi), %ymm0, %ymm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8> %a0, <32 x i8> %a1)
+  %1 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   %2 = load <32 x i8>, <32 x i8> *%a2, align 32
-  %3 = call <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8> %1, <32 x i8> %2)
+  %3 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> %1, <32 x i8> %2)
   ret <32 x i8> %3
 }
-declare <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <16 x i16> @test_psubsw(<16 x i16> %a0, <16 x i16> %a1, <16 x i16> *%a2) {
 ; GENERIC-LABEL: test_psubsw:
@@ -6548,12 +6548,12 @@ define <16 x i16> @test_psubsw(<16 x i16> %a0, <16 x i16> %a1, <16 x i16> *%a2)
 ; ZNVER1-NEXT:    vpsubsw %ymm1, %ymm0, %ymm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpsubsw (%rdi), %ymm0, %ymm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16> %a0, <16 x i16> %a1)
+  %1 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   %2 = load <16 x i16>, <16 x i16> *%a2, align 32
-  %3 = call <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16> %1, <16 x i16> %2)
+  %3 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> %1, <16 x i16> %2)
   ret <16 x i16> %3
 }
-declare <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <32 x i8> @test_psubusb(<32 x i8> %a0, <32 x i8> %a1, <32 x i8> *%a2) {
 ; GENERIC-LABEL: test_psubusb:

diff --git a/llvm/test/CodeGen/X86/sse2-schedule.ll b/llvm/test/CodeGen/X86/sse2-schedule.ll
index c895806..ca014b6 100644 (file)
--- a/llvm/test/CodeGen/X86/sse2-schedule.ll
+++ b/llvm/test/CodeGen/X86/sse2-schedule.ll
@@ -7934,12 +7934,12 @@ define <16 x i8> @test_paddsb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> *%a2) {
 ; ZNVER1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpaddsb (%rdi), %xmm0, %xmm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %a0, <16 x i8> %a1)
+  %1 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   %2 = load <16 x i8>, <16 x i8> *%a2, align 16
-  %3 = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %1, <16 x i8> %2)
+  %3 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %1, <16 x i8> %2)
   ret <16 x i8> %3
 }
-declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @test_paddsw(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> *%a2) {
 ; GENERIC-LABEL: test_paddsw:
@@ -8059,12 +8059,12 @@ define <8 x i16> @test_paddsw(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> *%a2) {
 ; ZNVER1-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpaddsw (%rdi), %xmm0, %xmm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %a0, <8 x i16> %a1)
+  %1 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   %2 = load <8 x i16>, <8 x i16> *%a2, align 16
-  %3 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %1, <8 x i16> %2)
+  %3 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %1, <8 x i16> %2)
   ret <8 x i16> %3
 }
-declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @test_paddusb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> *%a2) {
 ; GENERIC-LABEL: test_paddusb:
@@ -13870,12 +13870,12 @@ define <16 x i8> @test_psubsb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> *%a2) {
 ; ZNVER1-NEXT:    vpsubsb %xmm1, %xmm0, %xmm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpsubsb (%rdi), %xmm0, %xmm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %a0, <16 x i8> %a1)
+  %1 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   %2 = load <16 x i8>, <16 x i8> *%a2, align 16
-  %3 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %1, <16 x i8> %2)
+  %3 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %1, <16 x i8> %2)
   ret <16 x i8> %3
 }
-declare <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @test_psubsw(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> *%a2) {
 ; GENERIC-LABEL: test_psubsw:
@@ -13995,12 +13995,12 @@ define <8 x i16> @test_psubsw(<8 x i16> %a0, <8 x i16> %a1, <8 x i16> *%a2) {
 ; ZNVER1-NEXT:    vpsubsw %xmm1, %xmm0, %xmm0 # sched: [1:0.25]
 ; ZNVER1-NEXT:    vpsubsw (%rdi), %xmm0, %xmm0 # sched: [8:0.50]
 ; ZNVER1-NEXT:    retq # sched: [1:0.50]
-  %1 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %a0, <8 x i16> %a1)
+  %1 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   %2 = load <8 x i16>, <8 x i16> *%a2, align 16
-  %3 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %1, <8 x i16> %2)
+  %3 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %1, <8 x i16> %2)
   ret <8 x i16> %3
 }
-declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @test_psubusb(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> *%a2) {
 ; GENERIC-LABEL: test_psubusb:

diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx1.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx1.ll
index 6b9ed32..c694fc1 100644 (file)
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx1.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx1.ll
@@ -205,19 +205,19 @@ define <16 x i8> @stack_fold_paddsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddsb
   ;CHECK:       vpaddsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_paddsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_paddsw
   ;CHECK:       vpaddsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_paddusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddusb
@@ -930,19 +930,19 @@ define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubsb
   ;CHECK:       vpsubsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_psubsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_psubsw
   ;CHECK:       vpsubsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_psubusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubusb

diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
index 3cf3fb3..a961798 100644 (file)
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx2.ll
@@ -161,19 +161,19 @@ define <32 x i8> @stack_fold_paddsb(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddsb
   ;CHECK:       vpaddsb {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8> %a0, <32 x i8> %a1)
+  %2 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   ret <32 x i8> %2
 }
-declare <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <16 x i16> @stack_fold_paddsw(<16 x i16> %a0, <16 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_paddsw
   ;CHECK:       vpaddsw {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16> %a0, <16 x i16> %a1)
+  %2 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   ret <16 x i16> %2
 }
-declare <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <32 x i8> @stack_fold_paddusb(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddusb
@@ -1109,19 +1109,19 @@ define <32 x i8> @stack_fold_psubsb(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubsb
   ;CHECK:       vpsubsb {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8> %a0, <32 x i8> %a1)
+  %2 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   ret <32 x i8> %2
 }
-declare <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <16 x i16> @stack_fold_psubsw(<16 x i16> %a0, <16 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_psubsw
   ;CHECK:       vpsubsw {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16> %a0, <16 x i16> %a1)
+  %2 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   ret <16 x i16> %2
 }
-declare <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <32 x i8> @stack_fold_psubusb(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubusb

diff --git a/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll b/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
index f63fac1..f01bc61 100644 (file)
--- a/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-avx512vl.ll
@@ -421,37 +421,37 @@ define <16 x i8> @stack_fold_paddsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddsb
   ;CHECK:       vpaddsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <32 x i8> @stack_fold_paddsb_ymm(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddsb_ymm
   ;CHECK:       vpaddsb {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8> %a0, <32 x i8> %a1)
+  %2 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   ret <32 x i8> %2
 }
-declare <32 x i8> @llvm.x86.avx2.padds.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_paddsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_paddsw
   ;CHECK:       vpaddsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i16> @stack_fold_paddsw_ymm(<16 x i16> %a0, <16 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_paddsw_ymm
   ;CHECK:       vpaddsw {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16> %a0, <16 x i16> %a1)
+  %2 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   ret <16 x i16> %2
 }
-declare <16 x i16> @llvm.x86.avx2.padds.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_paddusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddusb
@@ -2355,37 +2355,37 @@ define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubsb
   ;CHECK:       vpsubsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <32 x i8> @stack_fold_psubsb_ymm(<32 x i8> %a0, <32 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubsb_ymm
   ;CHECK:       vpsubsb {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8> %a0, <32 x i8> %a1)
+  %2 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> %a0, <32 x i8> %a1)
   ret <32 x i8> %2
 }
-declare <32 x i8> @llvm.x86.avx2.psubs.b(<32 x i8>, <32 x i8>) nounwind readnone
+declare <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8>, <32 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_psubsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_psubsw
   ;CHECK:       vpsubsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}}, {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i16> @stack_fold_psubsw_ymm(<16 x i16> %a0, <16 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_psubsw_ymm
   ;CHECK:       vpsubsw {{-?[0-9]*}}(%rsp), {{%ymm[0-9][0-9]*}}, {{%ymm[0-9][0-9]*}} {{.*#+}} 32-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{xmm16},~{xmm17},~{xmm18},~{xmm19},~{xmm20},~{xmm21},~{xmm22},~{xmm23},~{xmm24},~{xmm25},~{xmm26},~{xmm27},~{xmm28},~{xmm29},~{xmm30},~{xmm31},~{flags}"()
-  %2 = call <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16> %a0, <16 x i16> %a1)
+  %2 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> %a0, <16 x i16> %a1)
   ret <16 x i16> %2
 }
-declare <16 x i16> @llvm.x86.avx2.psubs.w(<16 x i16>, <16 x i16>) nounwind readnone
+declare <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16>, <16 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_psubusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubusb

diff --git a/llvm/test/CodeGen/X86/stack-folding-int-sse42.ll b/llvm/test/CodeGen/X86/stack-folding-int-sse42.ll
index 374bdd6..c498e2b 100644 (file)
--- a/llvm/test/CodeGen/X86/stack-folding-int-sse42.ll
+++ b/llvm/test/CodeGen/X86/stack-folding-int-sse42.ll
@@ -241,19 +241,19 @@ define <16 x i8> @stack_fold_paddsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddsb
   ;CHECK:       paddsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.padds.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_paddsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_paddsw
   ;CHECK:       paddsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.padds.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_paddusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_paddusb
@@ -1084,19 +1084,19 @@ define <16 x i8> @stack_fold_psubsb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubsb
   ;CHECK:       psubsb {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8> %a0, <16 x i8> %a1)
+  %2 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> %a0, <16 x i8> %a1)
   ret <16 x i8> %2
 }
-declare <16 x i8> @llvm.x86.sse2.psubs.b(<16 x i8>, <16 x i8>) nounwind readnone
+declare <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
 
 define <8 x i16> @stack_fold_psubsw(<8 x i16> %a0, <8 x i16> %a1) {
   ;CHECK-LABEL: stack_fold_psubsw
   ;CHECK:       psubsw {{-?[0-9]*}}(%rsp), {{%xmm[0-9][0-9]*}} {{.*#+}} 16-byte Folded Reload
   %1 = tail call <2 x i64> asm sideeffect "nop", "=x,~{xmm2},~{xmm3},~{xmm4},~{xmm5},~{xmm6},~{xmm7},~{xmm8},~{xmm9},~{xmm10},~{xmm11},~{xmm12},~{xmm13},~{xmm14},~{xmm15},~{flags}"()
-  %2 = call <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16> %a0, <8 x i16> %a1)
+  %2 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> %a0, <8 x i16> %a1)
   ret <8 x i16> %2
 }
-declare <8 x i16> @llvm.x86.sse2.psubs.w(<8 x i16>, <8 x i16>) nounwind readnone
+declare <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
 
 define <16 x i8> @stack_fold_psubusb(<16 x i8> %a0, <16 x i8> %a1) {
   ;CHECK-LABEL: stack_fold_psubusb