[llvm] c960010 - [X86] isNegatibleForFree - allow pre-legalized FMA negation
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Fri Feb 7 09:04:43 PST 2020
Author: Simon Pilgrim
Date: 2020-02-07T17:04:17Z
New Revision: c96001035d4b7ac1b4b84705c7035d6c0fa28ffa
URL: https://github.com/llvm/llvm-project/commit/c96001035d4b7ac1b4b84705c7035d6c0fa28ffa
DIFF: https://github.com/llvm/llvm-project/commit/c96001035d4b7ac1b4b84705c7035d6c0fa28ffa.diff
LOG: [X86] isNegatibleForFree - allow pre-legalized FMA negation
As long as the FMA operation is legal (which we can proxy for the FMA3/FMA4 variants as well), we don't have to wait for the LegalOperations stage.
Added:
Modified:
llvm/lib/Target/X86/X86ISelLowering.cpp
llvm/test/CodeGen/X86/fma-fneg-combine-2.ll
llvm/test/CodeGen/X86/fma_patterns.ll
llvm/test/CodeGen/X86/fma_patterns_wide.ll
llvm/test/CodeGen/X86/recip-fastmath.ll
llvm/test/CodeGen/X86/recip-fastmath2.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 559804805ecf..c53639e98cf2 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -42926,7 +42926,8 @@ char X86TargetLowering::isNegatibleForFree(SDValue Op, SelectionDAG &DAG,
case X86ISD::FNMADD_RND:
case X86ISD::FNMSUB_RND: {
if (!Op.hasOneUse() || !Subtarget.hasAnyFMA() || !isTypeLegal(VT) ||
- !(SVT == MVT::f32 || SVT == MVT::f64) || !LegalOperations)
+ !(SVT == MVT::f32 || SVT == MVT::f64) ||
+ !isOperationLegal(ISD::FMA, VT))
break;
// This is always negatible for free but we might be able to remove some
@@ -42966,7 +42967,8 @@ SDValue X86TargetLowering::getNegatedExpression(SDValue Op, SelectionDAG &DAG,
case X86ISD::FNMADD_RND:
case X86ISD::FNMSUB_RND: {
if (!Op.hasOneUse() || !Subtarget.hasAnyFMA() || !isTypeLegal(VT) ||
- !(SVT == MVT::f32 || SVT == MVT::f64) || !LegalOperations)
+ !(SVT == MVT::f32 || SVT == MVT::f64) ||
+ !isOperationLegal(ISD::FMA, VT))
break;
// This is always negatible for free but we might be able to remove some
diff --git a/llvm/test/CodeGen/X86/fma-fneg-combine-2.ll b/llvm/test/CodeGen/X86/fma-fneg-combine-2.ll
index bc1e1beb427e..cf33c0c8c427 100644
--- a/llvm/test/CodeGen/X86/fma-fneg-combine-2.ll
+++ b/llvm/test/CodeGen/X86/fma-fneg-combine-2.ll
@@ -5,14 +5,14 @@
define float @test_fneg_fma_subx_y_negz_f32(float %w, float %x, float %y, float %z) {
; FMA3-LABEL: test_fneg_fma_subx_y_negz_f32:
; FMA3: # %bb.0: # %entry
-; FMA3-NEXT: vsubss %xmm0, %xmm1, %xmm0
-; FMA3-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; FMA3-NEXT: vsubss %xmm1, %xmm0, %xmm0
+; FMA3-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; FMA3-NEXT: retq
;
; FMA4-LABEL: test_fneg_fma_subx_y_negz_f32:
; FMA4: # %bb.0: # %entry
-; FMA4-NEXT: vsubss %xmm0, %xmm1, %xmm0
-; FMA4-NEXT: vfmaddss %xmm3, %xmm2, %xmm0, %xmm0
+; FMA4-NEXT: vsubss %xmm1, %xmm0, %xmm0
+; FMA4-NEXT: vfnmaddss %xmm3, %xmm2, %xmm0, %xmm0
; FMA4-NEXT: retq
entry:
%subx = fsub nsz float %w, %x
@@ -25,14 +25,14 @@ entry:
define float @test_fneg_fma_x_suby_negz_f32(float %w, float %x, float %y, float %z) {
; FMA3-LABEL: test_fneg_fma_x_suby_negz_f32:
; FMA3: # %bb.0: # %entry
-; FMA3-NEXT: vsubss %xmm0, %xmm2, %xmm0
-; FMA3-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; FMA3-NEXT: vsubss %xmm2, %xmm0, %xmm0
+; FMA3-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; FMA3-NEXT: retq
;
; FMA4-LABEL: test_fneg_fma_x_suby_negz_f32:
; FMA4: # %bb.0: # %entry
-; FMA4-NEXT: vsubss %xmm0, %xmm2, %xmm0
-; FMA4-NEXT: vfmaddss %xmm3, %xmm0, %xmm1, %xmm0
+; FMA4-NEXT: vsubss %xmm2, %xmm0, %xmm0
+; FMA4-NEXT: vfnmaddss %xmm3, %xmm0, %xmm1, %xmm0
; FMA4-NEXT: retq
entry:
%suby = fsub nsz float %w, %y
@@ -46,15 +46,15 @@ define float @test_fneg_fma_subx_suby_negz_f32(float %w, float %x, float %y, flo
; FMA3-LABEL: test_fneg_fma_subx_suby_negz_f32:
; FMA3: # %bb.0: # %entry
; FMA3-NEXT: vsubss %xmm1, %xmm0, %xmm1
-; FMA3-NEXT: vsubss %xmm0, %xmm2, %xmm0
-; FMA3-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; FMA3-NEXT: vsubss %xmm2, %xmm0, %xmm0
+; FMA3-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; FMA3-NEXT: retq
;
; FMA4-LABEL: test_fneg_fma_subx_suby_negz_f32:
; FMA4: # %bb.0: # %entry
; FMA4-NEXT: vsubss %xmm1, %xmm0, %xmm1
-; FMA4-NEXT: vsubss %xmm0, %xmm2, %xmm0
-; FMA4-NEXT: vfmaddss %xmm3, %xmm0, %xmm1, %xmm0
+; FMA4-NEXT: vsubss %xmm2, %xmm0, %xmm0
+; FMA4-NEXT: vfnmaddss %xmm3, %xmm0, %xmm1, %xmm0
; FMA4-NEXT: retq
entry:
%subx = fsub nsz float %w, %x
@@ -92,13 +92,13 @@ define float @negated_constant(float %x) {
; FMA3-LABEL: negated_constant:
; FMA3: # %bb.0:
; FMA3-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
-; FMA3-NEXT: vfmadd132ss {{.*#+}} xmm0 = (xmm0 * mem) + xmm1
+; FMA3-NEXT: vfnmsub132ss {{.*#+}} xmm0 = -(xmm0 * mem) - xmm1
; FMA3-NEXT: retq
;
; FMA4-LABEL: negated_constant:
; FMA4: # %bb.0:
; FMA4-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
-; FMA4-NEXT: vfmaddss %xmm1, {{.*}}(%rip), %xmm0, %xmm0
+; FMA4-NEXT: vfnmsubss %xmm1, {{.*}}(%rip), %xmm0, %xmm0
; FMA4-NEXT: retq
%m = fmul float %x, 42.0
%fma = call nsz float @llvm.fma.f32(float %x, float -42.0, float %m)
diff --git a/llvm/test/CodeGen/X86/fma_patterns.ll b/llvm/test/CodeGen/X86/fma_patterns.ll
index a62ac70c4f7e..043a5206f296 100644
--- a/llvm/test/CodeGen/X86/fma_patterns.ll
+++ b/llvm/test/CodeGen/X86/fma_patterns.ll
@@ -1293,20 +1293,20 @@ define float @test_f32_interp(float %x, float %y, float %t) {
;
; FMA-NOINFS-LABEL: test_f32_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213ss {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; FMA-NOINFS-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; FMA-NOINFS-NEXT: vfmsub213ss {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; FMA-NOINFS-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_f32_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddss %xmm1, %xmm1, %xmm2, %xmm1
-; FMA4-NOINFS-NEXT: vfmaddss %xmm1, %xmm2, %xmm0, %xmm0
+; FMA4-NOINFS-NEXT: vfmsubss %xmm1, %xmm1, %xmm2, %xmm1
+; FMA4-NOINFS-NEXT: vfmsubss %xmm1, %xmm2, %xmm0, %xmm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_f32_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213ss {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; AVX512-NOINFS-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; AVX512-NOINFS-NEXT: vfmsub213ss {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; AVX512-NOINFS-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub float 1.0, %t
%tx = fmul float %x, %t
@@ -1342,20 +1342,20 @@ define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float
;
; FMA-NOINFS-LABEL: test_v4f32_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213ps {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; FMA-NOINFS-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v4f32_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddps %xmm1, %xmm1, %xmm2, %xmm1
-; FMA4-NOINFS-NEXT: vfmaddps %xmm1, %xmm2, %xmm0, %xmm0
+; FMA4-NOINFS-NEXT: vfmsubps %xmm1, %xmm1, %xmm2, %xmm1
+; FMA4-NOINFS-NEXT: vfmsubps %xmm1, %xmm2, %xmm0, %xmm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v4f32_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213ps {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; AVX512-NOINFS-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %t
%tx = fmul <4 x float> %x, %t
@@ -1391,20 +1391,20 @@ define <8 x float> @test_v8f32_interp(<8 x float> %x, <8 x float> %y, <8 x float
;
; FMA-NOINFS-LABEL: test_v8f32_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm1
-; FMA-NOINFS-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm1
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v8f32_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddps %ymm1, %ymm1, %ymm2, %ymm1
-; FMA4-NOINFS-NEXT: vfmaddps %ymm1, %ymm2, %ymm0, %ymm0
+; FMA4-NOINFS-NEXT: vfmsubps %ymm1, %ymm1, %ymm2, %ymm1
+; FMA4-NOINFS-NEXT: vfmsubps %ymm1, %ymm2, %ymm0, %ymm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v8f32_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm1
-; AVX512-NOINFS-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <8 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %t
%tx = fmul <8 x float> %x, %t
@@ -1440,20 +1440,20 @@ define double @test_f64_interp(double %x, double %y, double %t) {
;
; FMA-NOINFS-LABEL: test_f64_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213sd {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; FMA-NOINFS-NEXT: vfmadd213sd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; FMA-NOINFS-NEXT: vfmsub213sd {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; FMA-NOINFS-NEXT: vfmsub213sd {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_f64_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddsd %xmm1, %xmm1, %xmm2, %xmm1
-; FMA4-NOINFS-NEXT: vfmaddsd %xmm1, %xmm2, %xmm0, %xmm0
+; FMA4-NOINFS-NEXT: vfmsubsd %xmm1, %xmm1, %xmm2, %xmm1
+; FMA4-NOINFS-NEXT: vfmsubsd %xmm1, %xmm2, %xmm0, %xmm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_f64_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213sd {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; AVX512-NOINFS-NEXT: vfmadd213sd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; AVX512-NOINFS-NEXT: vfmsub213sd {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; AVX512-NOINFS-NEXT: vfmsub213sd {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub double 1.0, %t
%tx = fmul double %x, %t
@@ -1489,20 +1489,20 @@ define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x do
;
; FMA-NOINFS-LABEL: test_v2f64_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213pd {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; FMA-NOINFS-NEXT: vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v2f64_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddpd %xmm1, %xmm1, %xmm2, %xmm1
-; FMA4-NOINFS-NEXT: vfmaddpd %xmm1, %xmm2, %xmm0, %xmm0
+; FMA4-NOINFS-NEXT: vfmsubpd %xmm1, %xmm1, %xmm2, %xmm1
+; FMA4-NOINFS-NEXT: vfmsubpd %xmm1, %xmm2, %xmm0, %xmm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v2f64_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213pd {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
-; AVX512-NOINFS-NEXT: vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} xmm1 = (xmm2 * xmm1) - xmm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} xmm0 = (xmm2 * xmm0) - xmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <2 x double> <double 1.0, double 1.0>, %t
%tx = fmul <2 x double> %x, %t
@@ -1538,20 +1538,20 @@ define <4 x double> @test_v4f64_interp(<4 x double> %x, <4 x double> %y, <4 x do
;
; FMA-NOINFS-LABEL: test_v4f64_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213pd {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm1
-; FMA-NOINFS-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm1
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm1
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v4f64_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddpd %ymm1, %ymm1, %ymm2, %ymm1
-; FMA4-NOINFS-NEXT: vfmaddpd %ymm1, %ymm2, %ymm0, %ymm0
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm1, %ymm1, %ymm2, %ymm1
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm1, %ymm2, %ymm0, %ymm0
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v4f64_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213pd {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm1
-; AVX512-NOINFS-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <4 x double> <double 1.0, double 1.0, double 1.0, double 1.0>, %t
%tx = fmul <4 x double> %x, %t
diff --git a/llvm/test/CodeGen/X86/fma_patterns_wide.ll b/llvm/test/CodeGen/X86/fma_patterns_wide.ll
index 98807d33db04..773352b6609c 100644
--- a/llvm/test/CodeGen/X86/fma_patterns_wide.ll
+++ b/llvm/test/CodeGen/X86/fma_patterns_wide.ll
@@ -849,24 +849,24 @@ define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x
;
; FMA-NOINFS-LABEL: test_v16f32_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm5 * ymm3) + ymm3
-; FMA-NOINFS-NEXT: vfnmadd213ps {{.*#+}} ymm2 = -(ymm4 * ymm2) + ymm2
-; FMA-NOINFS-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm2
-; FMA-NOINFS-NEXT: vfmadd213ps {{.*#+}} ymm1 = (ymm5 * ymm1) + ymm3
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm3 = (ymm5 * ymm3) - ymm3
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm2 = (ymm4 * ymm2) - ymm2
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm2
+; FMA-NOINFS-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm5 * ymm1) - ymm3
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v16f32_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddps %ymm3, %ymm3, %ymm5, %ymm3
-; FMA4-NOINFS-NEXT: vfnmaddps %ymm2, %ymm2, %ymm4, %ymm2
-; FMA4-NOINFS-NEXT: vfmaddps %ymm2, %ymm4, %ymm0, %ymm0
-; FMA4-NOINFS-NEXT: vfmaddps %ymm3, %ymm5, %ymm1, %ymm1
+; FMA4-NOINFS-NEXT: vfmsubps %ymm3, %ymm3, %ymm5, %ymm3
+; FMA4-NOINFS-NEXT: vfmsubps %ymm2, %ymm2, %ymm4, %ymm2
+; FMA4-NOINFS-NEXT: vfmsubps %ymm2, %ymm4, %ymm0, %ymm0
+; FMA4-NOINFS-NEXT: vfmsubps %ymm3, %ymm5, %ymm1, %ymm1
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v16f32_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213ps {{.*#+}} zmm1 = -(zmm2 * zmm1) + zmm1
-; AVX512-NOINFS-NEXT: vfmadd213ps {{.*#+}} zmm0 = (zmm2 * zmm0) + zmm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} zmm1 = (zmm2 * zmm1) - zmm1
+; AVX512-NOINFS-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm2 * zmm0) - zmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <16 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %t
%tx = fmul <16 x float> %x, %t
@@ -908,24 +908,24 @@ define <8 x double> @test_v8f64_interp(<8 x double> %x, <8 x double> %y, <8 x do
;
; FMA-NOINFS-LABEL: test_v8f64_interp:
; FMA-NOINFS: # %bb.0:
-; FMA-NOINFS-NEXT: vfnmadd213pd {{.*#+}} ymm3 = -(ymm5 * ymm3) + ymm3
-; FMA-NOINFS-NEXT: vfnmadd213pd {{.*#+}} ymm2 = -(ymm4 * ymm2) + ymm2
-; FMA-NOINFS-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm2
-; FMA-NOINFS-NEXT: vfmadd213pd {{.*#+}} ymm1 = (ymm5 * ymm1) + ymm3
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm3 = (ymm5 * ymm3) - ymm3
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm2 = (ymm4 * ymm2) - ymm2
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm2
+; FMA-NOINFS-NEXT: vfmsub213pd {{.*#+}} ymm1 = (ymm5 * ymm1) - ymm3
; FMA-NOINFS-NEXT: retq
;
; FMA4-NOINFS-LABEL: test_v8f64_interp:
; FMA4-NOINFS: # %bb.0:
-; FMA4-NOINFS-NEXT: vfnmaddpd %ymm3, %ymm3, %ymm5, %ymm3
-; FMA4-NOINFS-NEXT: vfnmaddpd %ymm2, %ymm2, %ymm4, %ymm2
-; FMA4-NOINFS-NEXT: vfmaddpd %ymm2, %ymm4, %ymm0, %ymm0
-; FMA4-NOINFS-NEXT: vfmaddpd %ymm3, %ymm5, %ymm1, %ymm1
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm3, %ymm3, %ymm5, %ymm3
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm2, %ymm2, %ymm4, %ymm2
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm2, %ymm4, %ymm0, %ymm0
+; FMA4-NOINFS-NEXT: vfmsubpd %ymm3, %ymm5, %ymm1, %ymm1
; FMA4-NOINFS-NEXT: retq
;
; AVX512-NOINFS-LABEL: test_v8f64_interp:
; AVX512-NOINFS: # %bb.0:
-; AVX512-NOINFS-NEXT: vfnmadd213pd {{.*#+}} zmm1 = -(zmm2 * zmm1) + zmm1
-; AVX512-NOINFS-NEXT: vfmadd213pd {{.*#+}} zmm0 = (zmm2 * zmm0) + zmm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} zmm1 = (zmm2 * zmm1) - zmm1
+; AVX512-NOINFS-NEXT: vfmsub213pd {{.*#+}} zmm0 = (zmm2 * zmm0) - zmm1
; AVX512-NOINFS-NEXT: retq
%t1 = fsub <8 x double> <double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0>, %t
%tx = fmul <8 x double> %x, %t
diff --git a/llvm/test/CodeGen/X86/recip-fastmath.ll b/llvm/test/CodeGen/X86/recip-fastmath.ll
index 5d05bd401e24..08d59aacfd39 100644
--- a/llvm/test/CodeGen/X86/recip-fastmath.ll
+++ b/llvm/test/CodeGen/X86/recip-fastmath.ll
@@ -60,14 +60,14 @@ define float @f32_one_step(float %x) #1 {
; FMA-RECIP-LABEL: f32_one_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; FMA-RECIP-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; FMA-RECIP-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: f32_one_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; BDVER2-NEXT: vfmaddss {{.*}}(%rip), %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfmsubss {{.*}}(%rip), %xmm1, %xmm0, %xmm0
; BDVER2-NEXT: vfnmaddss %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
@@ -94,7 +94,7 @@ define float @f32_one_step(float %x) #1 {
; HASWELL-LABEL: f32_one_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; HASWELL-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; HASWELL-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; HASWELL-NEXT: retq
;
@@ -111,7 +111,7 @@ define float @f32_one_step(float %x) #1 {
; AVX512-LABEL: f32_one_step:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; AVX512-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; AVX512-NEXT: retq
%div = fdiv fast float 1.0, %x
@@ -144,8 +144,8 @@ define float @f32_one_step_variables(float %x, float %y) #1 {
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpss %xmm1, %xmm1, %xmm2
; FMA-RECIP-NEXT: vmulss %xmm2, %xmm0, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; FMA-RECIP-NEXT: vfmadd213ss {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; FMA-RECIP-NEXT: vmovaps %xmm2, %xmm0
; FMA-RECIP-NEXT: retq
;
@@ -153,8 +153,8 @@ define float @f32_one_step_variables(float %x, float %y) #1 {
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpss %xmm1, %xmm1, %xmm2
; BDVER2-NEXT: vmulss %xmm2, %xmm0, %xmm3
-; BDVER2-NEXT: vfnmaddss %xmm0, %xmm3, %xmm1, %xmm0
-; BDVER2-NEXT: vfmaddss %xmm3, %xmm0, %xmm2, %xmm0
+; BDVER2-NEXT: vfmsubss %xmm0, %xmm3, %xmm1, %xmm0
+; BDVER2-NEXT: vfnmaddss %xmm3, %xmm0, %xmm2, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: f32_one_step_variables:
@@ -181,8 +181,8 @@ define float @f32_one_step_variables(float %x, float %y) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpss %xmm1, %xmm1, %xmm2
; HASWELL-NEXT: vmulss %xmm2, %xmm0, %xmm3
-; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; HASWELL-NEXT: vfmadd213ss {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; HASWELL-NEXT: vmovaps %xmm2, %xmm0
; HASWELL-NEXT: retq
;
@@ -200,8 +200,8 @@ define float @f32_one_step_variables(float %x, float %y) #1 {
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpss %xmm1, %xmm1, %xmm2
; AVX512-NEXT: vmulss %xmm2, %xmm0, %xmm3
-; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; AVX512-NEXT: vmovaps %xmm2, %xmm0
; AVX512-NEXT: retq
%div = fdiv fast float %x, %y
@@ -245,20 +245,20 @@ define float @f32_two_step(float %x) #2 {
; FMA-RECIP-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; FMA-RECIP-NEXT: vmovaps %xmm1, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; FMA-RECIP-NEXT: vfmadd132ss {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; FMA-RECIP-NEXT: vfmadd132ss {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; FMA-RECIP-NEXT: vfnmadd132ss {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; FMA-RECIP-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; BDVER2-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; BDVER2-NEXT: vfnmaddss %xmm2, %xmm1, %xmm0, %xmm3
-; BDVER2-NEXT: vfmaddss %xmm1, %xmm3, %xmm1, %xmm1
-; BDVER2-NEXT: vfnmaddss %xmm2, %xmm1, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddss %xmm1, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubss %xmm2, %xmm1, %xmm0, %xmm3
+; BDVER2-NEXT: vfnmaddss %xmm1, %xmm3, %xmm1, %xmm1
+; BDVER2-NEXT: vfmsubss %xmm2, %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddss %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: f32_two_step:
@@ -294,10 +294,10 @@ define float @f32_two_step(float %x) #2 {
; HASWELL-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; HASWELL-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; HASWELL-NEXT: vmovaps %xmm1, %xmm3
-; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; HASWELL-NEXT: vfmadd132ss {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd132ss {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; HASWELL-NEXT: vfnmadd132ss {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: f32_two_step:
@@ -319,10 +319,10 @@ define float @f32_two_step(float %x) #2 {
; AVX512-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; AVX512-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; AVX512-NEXT: vmovaps %xmm1, %xmm3
-; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; AVX512-NEXT: vfmadd132ss {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; AVX512-NEXT: vfmadd132ss {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; AVX512-NEXT: vfnmadd132ss {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; AVX512-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast float 1.0, %x
ret float %div
@@ -412,15 +412,15 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; FMA-RECIP-LABEL: v4f32_one_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v4f32_one_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1
-; BDVER2-NEXT: vfnmaddps {{.*}}(%rip), %xmm1, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm1, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v4f32_one_step:
@@ -447,8 +447,8 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v4f32_one_step:
@@ -465,15 +465,15 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
-; KNL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; KNL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - xmm2
+; KNL-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; KNL-NEXT: retq
;
; SKX-LABEL: v4f32_one_step:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %xmm0, %xmm1
-; SKX-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
-; SKX-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; SKX-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
+; SKX-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; SKX-NEXT: retq
%div = fdiv fast <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <4 x float> %div
@@ -505,8 +505,8 @@ define <4 x float> @v4f32_one_step_variables(<4 x float> %x, <4 x float> %y) #1
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm1, %xmm2
; FMA-RECIP-NEXT: vmulps %xmm2, %xmm0, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; FMA-RECIP-NEXT: vmovaps %xmm2, %xmm0
; FMA-RECIP-NEXT: retq
;
@@ -514,8 +514,8 @@ define <4 x float> @v4f32_one_step_variables(<4 x float> %x, <4 x float> %y) #1
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm1, %xmm2
; BDVER2-NEXT: vmulps %xmm2, %xmm0, %xmm3
-; BDVER2-NEXT: vfnmaddps %xmm0, %xmm3, %xmm1, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm3, %xmm0, %xmm2, %xmm0
+; BDVER2-NEXT: vfmsubps %xmm0, %xmm3, %xmm1, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm3, %xmm0, %xmm2, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v4f32_one_step_variables:
@@ -542,8 +542,8 @@ define <4 x float> @v4f32_one_step_variables(<4 x float> %x, <4 x float> %y) #1
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm1, %xmm2
; HASWELL-NEXT: vmulps %xmm2, %xmm0, %xmm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; HASWELL-NEXT: vmovaps %xmm2, %xmm0
; HASWELL-NEXT: retq
;
@@ -561,8 +561,8 @@ define <4 x float> @v4f32_one_step_variables(<4 x float> %x, <4 x float> %y) #1
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpps %xmm1, %xmm2
; AVX512-NEXT: vmulps %xmm2, %xmm0, %xmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm1 = -(xmm3 * xmm1) + xmm0
-; AVX512-NEXT: vfmadd213ps {{.*#+}} xmm2 = (xmm1 * xmm2) + xmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} xmm1 = (xmm3 * xmm1) - xmm0
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm2 = -(xmm1 * xmm2) + xmm3
; AVX512-NEXT: vmovaps %xmm2, %xmm0
; AVX512-NEXT: retq
%div = fdiv fast <4 x float> %x, %y
@@ -606,20 +606,20 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-RECIP-NEXT: vmovaps %xmm1, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v4f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm3
-; BDVER2-NEXT: vfmaddps %xmm1, %xmm3, %xmm1, %xmm1
-; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm1, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubps %xmm2, %xmm1, %xmm0, %xmm3
+; BDVER2-NEXT: vfnmaddps %xmm1, %xmm3, %xmm1, %xmm1
+; BDVER2-NEXT: vfmsubps %xmm2, %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v4f32_two_step:
@@ -655,10 +655,10 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; HASWELL-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NEXT: vmovaps %xmm1, %xmm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v4f32_two_step:
@@ -680,10 +680,10 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; AVX512-NEXT: vrcpps %xmm0, %xmm1
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-NEXT: vmovaps %xmm1, %xmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} xmm3 = (xmm0 * xmm3) - xmm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} xmm3 = -(xmm3 * xmm1) + xmm1
+; AVX512-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm3) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <4 x float> %div
@@ -783,15 +783,15 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; FMA-RECIP-LABEL: v8f32_one_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - mem
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v8f32_one_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1
-; BDVER2-NEXT: vfnmaddps {{.*}}(%rip), %ymm1, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm1, %ymm0, %ymm1, %ymm0
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %ymm1, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm1, %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v8f32_one_step:
@@ -818,8 +818,8 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - ymm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v8f32_one_step:
@@ -836,15 +836,15 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2
-; KNL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; KNL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - ymm2
+; KNL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; KNL-NEXT: retq
;
; SKX-LABEL: v8f32_one_step:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %ymm0, %ymm1
-; SKX-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
-; SKX-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; SKX-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - mem
+; SKX-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; SKX-NEXT: retq
%div = fdiv fast <8 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <8 x float> %div
@@ -900,20 +900,20 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-RECIP-NEXT: vmovaps %ymm1, %ymm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm3) + ymm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm3 = (ymm0 * ymm3) - ymm2
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm3 = -(ymm3 * ymm1) + ymm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm3) + ymm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v8f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1
; BDVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm3
-; BDVER2-NEXT: vfmaddps %ymm1, %ymm3, %ymm1, %ymm1
-; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm1, %ymm0, %ymm1, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm2, %ymm1, %ymm0, %ymm3
+; BDVER2-NEXT: vfnmaddps %ymm1, %ymm3, %ymm1, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm2, %ymm1, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm1, %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v8f32_two_step:
@@ -949,10 +949,10 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; HASWELL-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NEXT: vmovaps %ymm1, %ymm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm3) + ymm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm3 = (ymm0 * ymm3) - ymm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm3 = -(ymm3 * ymm1) + ymm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm3) + ymm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v8f32_two_step:
@@ -974,10 +974,10 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; AVX512-NEXT: vrcpps %ymm0, %ymm1
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-NEXT: vmovaps %ymm1, %ymm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm3) + ymm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} ymm3 = (ymm0 * ymm3) - ymm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} ymm3 = -(ymm3 * ymm1) + ymm1
+; AVX512-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm3) + ymm3
; AVX512-NEXT: retq
%div = fdiv fast <8 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <8 x float> %div
@@ -1110,11 +1110,11 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm2) + ymm2
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm2) + ymm2
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v16f32_one_step:
@@ -1122,10 +1122,10 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; BDVER2-NEXT: vrcpps %ymm0, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; BDVER2-NEXT: vrcpps %ymm1, %ymm4
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm4, %ymm1, %ymm1
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm0, %ymm2, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm4, %ymm1, %ymm4, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm0, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm4, %ymm1, %ymm1
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm0, %ymm2, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm4, %ymm1, %ymm4, %ymm1
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v16f32_one_step:
@@ -1163,10 +1163,10 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; HASWELL-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NEXT: vrcpps %ymm1, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm4) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm2) + ymm2
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm4) + ymm4
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v16f32_one_step:
@@ -1187,8 +1187,8 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; AVX512-LABEL: v16f32_one_step:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcp14ps %zmm0, %zmm1
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm1 * zmm0) + mem
-; AVX512-NEXT: vfmadd132ps {{.*#+}} zmm0 = (zmm0 * zmm1) + zmm1
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm1 * zmm0) - mem
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} zmm0 = -(zmm0 * zmm1) + zmm1
; AVX512-NEXT: retq
%div = fdiv fast <16 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <16 x float> %div
@@ -1279,31 +1279,31 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-RECIP-NEXT: vmovaps %ymm2, %ymm4
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm4) + ymm4
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm0 * ymm4) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm4) + ymm4
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
; FMA-RECIP-NEXT: vmovaps %ymm2, %ymm4
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm1 * ymm4) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm4) + ymm4
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm1 * ymm4) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm4) + ymm4
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v16f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm4
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm4, %ymm2, %ymm2
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm0, %ymm2, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm0, %ymm4
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm4, %ymm2, %ymm2
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm0, %ymm2, %ymm0
; BDVER2-NEXT: vrcpps %ymm1, %ymm2
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm4
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm4, %ymm2, %ymm2
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm1
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm1, %ymm2, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm1, %ymm4
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm4, %ymm2, %ymm2
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm1, %ymm1
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm2, %ymm1
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v16f32_two_step:
@@ -1357,16 +1357,16 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; HASWELL-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NEXT: vmovaps %ymm2, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm4) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm0 * ymm4) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm4) + ymm4
; HASWELL-NEXT: vrcpps %ymm1, %ymm2
; HASWELL-NEXT: vmovaps %ymm2, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm1 * ymm4) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm4) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm1 * ymm4) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm4) + ymm4
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v16f32_two_step:
@@ -1397,10 +1397,10 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; AVX512-NEXT: vrcp14ps %zmm0, %zmm1
; AVX512-NEXT: vbroadcastss {{.*#+}} zmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-NEXT: vmovaps %zmm1, %zmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm3 = -(zmm0 * zmm3) + zmm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} zmm3 = (zmm3 * zmm1) + zmm1
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm3 * zmm0) + zmm2
-; AVX512-NEXT: vfmadd132ps {{.*#+}} zmm0 = (zmm0 * zmm3) + zmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm3 = (zmm0 * zmm3) - zmm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} zmm3 = -(zmm3 * zmm1) + zmm1
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm3 * zmm0) - zmm2
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} zmm0 = -(zmm0 * zmm3) + zmm3
; AVX512-NEXT: retq
%div = fdiv fast <16 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
ret <16 x float> %div
diff --git a/llvm/test/CodeGen/X86/recip-fastmath2.ll b/llvm/test/CodeGen/X86/recip-fastmath2.ll
index c5e364280b80..68bd3e92f5bd 100644
--- a/llvm/test/CodeGen/X86/recip-fastmath2.ll
+++ b/llvm/test/CodeGen/X86/recip-fastmath2.ll
@@ -58,8 +58,8 @@ define float @f32_one_step_2(float %x) #1 {
; FMA-RECIP-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; FMA-RECIP-NEXT: vmulss %xmm2, %xmm1, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; FMA-RECIP-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: f32_one_step_2:
@@ -67,8 +67,8 @@ define float @f32_one_step_2(float %x) #1 {
; BDVER2-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; BDVER2-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; BDVER2-NEXT: vmulss %xmm2, %xmm1, %xmm3
-; BDVER2-NEXT: vfnmaddss %xmm2, %xmm3, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddss %xmm3, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubss %xmm2, %xmm3, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddss %xmm3, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: f32_one_step_2:
@@ -98,8 +98,8 @@ define float @f32_one_step_2(float %x) #1 {
; HASWELL-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; HASWELL-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; HASWELL-NEXT: vmulss %xmm2, %xmm1, %xmm3
-; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: f32_one_step_2:
@@ -118,8 +118,8 @@ define float @f32_one_step_2(float %x) #1 {
; AVX512-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; AVX512-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; AVX512-NEXT: vmulss %xmm2, %xmm1, %xmm3
-; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast float 3456.0, %x
ret float %div
@@ -154,7 +154,7 @@ define float @f32_one_step_2_divs(float %x) #1 {
; FMA-RECIP-LABEL: f32_one_step_2_divs:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; FMA-RECIP-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; FMA-RECIP-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; FMA-RECIP-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
; FMA-RECIP-NEXT: vmulss %xmm0, %xmm1, %xmm0
@@ -163,7 +163,7 @@ define float @f32_one_step_2_divs(float %x) #1 {
; BDVER2-LABEL: f32_one_step_2_divs:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; BDVER2-NEXT: vfmaddss {{.*}}(%rip), %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfmsubss {{.*}}(%rip), %xmm1, %xmm0, %xmm0
; BDVER2-NEXT: vfnmaddss %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
; BDVER2-NEXT: vmulss %xmm0, %xmm1, %xmm0
@@ -196,7 +196,7 @@ define float @f32_one_step_2_divs(float %x) #1 {
; HASWELL-LABEL: f32_one_step_2_divs:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; HASWELL-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; HASWELL-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; HASWELL-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
; HASWELL-NEXT: vmulss %xmm0, %xmm1, %xmm0
@@ -217,7 +217,7 @@ define float @f32_one_step_2_divs(float %x) #1 {
; AVX512-LABEL: f32_one_step_2_divs:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
; AVX512-NEXT: vfnmadd132ss {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; AVX512-NEXT: vmulss {{.*}}(%rip), %xmm0, %xmm1
; AVX512-NEXT: vmulss %xmm0, %xmm1, %xmm0
@@ -267,23 +267,23 @@ define float @f32_two_step_2(float %x) #2 {
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; FMA-RECIP-NEXT: vfmadd231ss {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; FMA-RECIP-NEXT: vfmsub231ss {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; FMA-RECIP-NEXT: vfnmadd132ss {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; FMA-RECIP-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; FMA-RECIP-NEXT: vmulss %xmm1, %xmm2, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; FMA-RECIP-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; FMA-RECIP-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: f32_two_step_2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpss %xmm0, %xmm0, %xmm1
-; BDVER2-NEXT: vfmaddss {{.*}}(%rip), %xmm1, %xmm0, %xmm2
+; BDVER2-NEXT: vfmsubss {{.*}}(%rip), %xmm1, %xmm0, %xmm2
; BDVER2-NEXT: vmovss {{.*#+}} xmm4 = mem[0],zero,zero,zero
; BDVER2-NEXT: vfnmaddss %xmm1, %xmm2, %xmm1, %xmm1
; BDVER2-NEXT: vmulss %xmm4, %xmm1, %xmm3
-; BDVER2-NEXT: vfnmaddss %xmm4, %xmm3, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddss %xmm3, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubss %xmm4, %xmm3, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddss %xmm3, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: f32_two_step_2:
@@ -322,12 +322,12 @@ define float @f32_two_step_2(float %x) #2 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; HASWELL-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; HASWELL-NEXT: vfmadd231ss {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; HASWELL-NEXT: vfmsub231ss {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; HASWELL-NEXT: vfnmadd132ss {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; HASWELL-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; HASWELL-NEXT: vmulss %xmm1, %xmm2, %xmm3
-; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; HASWELL-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; HASWELL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; HASWELL-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: f32_two_step_2:
@@ -350,12 +350,12 @@ define float @f32_two_step_2(float %x) #2 {
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpss %xmm0, %xmm0, %xmm1
; AVX512-NEXT: vmovss {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; AVX512-NEXT: vfmadd231ss {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; AVX512-NEXT: vfmsub231ss {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; AVX512-NEXT: vfnmadd132ss {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; AVX512-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; AVX512-NEXT: vmulss %xmm1, %xmm2, %xmm3
-; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; AVX512-NEXT: vfmadd213ss {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; AVX512-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; AVX512-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast float 6789.0, %x
ret float %div
@@ -391,8 +391,8 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; FMA-RECIP-NEXT: vmulps %xmm2, %xmm1, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v4f32_one_step2:
@@ -400,8 +400,8 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; BDVER2-NEXT: vrcpps %xmm0, %xmm1
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; BDVER2-NEXT: vmulps %xmm2, %xmm1, %xmm3
-; BDVER2-NEXT: vfnmaddps %xmm2, %xmm3, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm3, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubps %xmm2, %xmm3, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm3, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v4f32_one_step2:
@@ -431,8 +431,8 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; HASWELL-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; HASWELL-NEXT: vmulps %xmm2, %xmm1, %xmm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v4f32_one_step2:
@@ -451,8 +451,8 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; AVX512-NEXT: vrcpps %xmm0, %xmm1
; AVX512-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; AVX512-NEXT: vmulps %xmm2, %xmm1, %xmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
-; AVX512-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm2
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast <4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, %x
ret <4 x float> %div
@@ -487,8 +487,8 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; FMA-RECIP-LABEL: v4f32_one_step_2_divs:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; FMA-RECIP-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1
; FMA-RECIP-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA-RECIP-NEXT: retq
@@ -496,8 +496,8 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; BDVER2-LABEL: v4f32_one_step_2_divs:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1
-; BDVER2-NEXT: vfnmaddps {{.*}}(%rip), %xmm1, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm1, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %xmm1, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm1, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1
; BDVER2-NEXT: vmulps %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
@@ -530,8 +530,8 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - xmm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; HASWELL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1
; HASWELL-NEXT: vmulps %xmm0, %xmm1, %xmm0
; HASWELL-NEXT: retq
@@ -552,8 +552,8 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
-; KNL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; KNL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - xmm2
+; KNL-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; KNL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1
; KNL-NEXT: vmulps %xmm0, %xmm1, %xmm0
; KNL-NEXT: retq
@@ -561,8 +561,8 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; SKX-LABEL: v4f32_one_step_2_divs:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %xmm0, %xmm1
-; SKX-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
-; SKX-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
+; SKX-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm1 * xmm0) - mem
+; SKX-NEXT: vfnmadd132ps {{.*#+}} xmm0 = -(xmm0 * xmm1) + xmm1
; SKX-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1
; SKX-NEXT: vmulps %xmm0, %xmm1, %xmm0
; SKX-NEXT: retq
@@ -610,24 +610,24 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; FMA-RECIP-LABEL: v4f32_two_step2:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; FMA-RECIP-NEXT: vfmadd231ps {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; FMA-RECIP-NEXT: vfmsub231ps {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; FMA-RECIP-NEXT: vmulps %xmm1, %xmm2, %xmm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v4f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1
-; BDVER2-NEXT: vfmaddps {{.*}}(%rip), %xmm1, %xmm0, %xmm2
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %xmm1, %xmm0, %xmm2
; BDVER2-NEXT: vmovaps {{.*#+}} xmm4 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; BDVER2-NEXT: vfnmaddps %xmm1, %xmm2, %xmm1, %xmm1
; BDVER2-NEXT: vmulps %xmm4, %xmm1, %xmm3
-; BDVER2-NEXT: vfnmaddps %xmm4, %xmm3, %xmm0, %xmm0
-; BDVER2-NEXT: vfmaddps %xmm3, %xmm0, %xmm1, %xmm0
+; BDVER2-NEXT: vfmsubps %xmm4, %xmm3, %xmm0, %xmm0
+; BDVER2-NEXT: vfnmaddps %xmm3, %xmm0, %xmm1, %xmm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v4f32_two_step2:
@@ -665,13 +665,13 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; HASWELL-LABEL: v4f32_two_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1
-; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; HASWELL-NEXT: vfmadd231ps {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; HASWELL-NEXT: vfmsub231ps {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; HASWELL-NEXT: vfnmadd132ps {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; HASWELL-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; HASWELL-NEXT: vmulps %xmm1, %xmm2, %xmm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v4f32_two_step2:
@@ -693,13 +693,13 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; AVX512-LABEL: v4f32_two_step2:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpps %xmm0, %xmm1
-; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; AVX512-NEXT: vfmadd231ps {{.*#+}} xmm2 = (xmm0 * xmm1) + xmm2
+; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; AVX512-NEXT: vfmsub231ps {{.*#+}} xmm2 = (xmm0 * xmm1) - xmm2
; AVX512-NEXT: vfnmadd132ps {{.*#+}} xmm2 = -(xmm2 * xmm1) + xmm1
; AVX512-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; AVX512-NEXT: vmulps %xmm1, %xmm2, %xmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm1
-; AVX512-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} xmm0 = (xmm3 * xmm0) - xmm1
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm2 * xmm0) + xmm3
; AVX512-NEXT: retq
%div = fdiv fast <4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, %x
ret <4 x float> %div
@@ -744,8 +744,8 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; FMA-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + ymm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v8f32_one_step2:
@@ -753,8 +753,8 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; BDVER2-NEXT: vrcpps %ymm0, %ymm1
; BDVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; BDVER2-NEXT: vmulps %ymm2, %ymm1, %ymm3
-; BDVER2-NEXT: vfnmaddps %ymm2, %ymm3, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm3, %ymm0, %ymm1, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm2, %ymm3, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm3, %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v8f32_one_step2:
@@ -784,8 +784,8 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; HASWELL-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; HASWELL-NEXT: vmulps %ymm2, %ymm1, %ymm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + ymm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v8f32_one_step2:
@@ -804,8 +804,8 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; AVX512-NEXT: vrcpps %ymm0, %ymm1
; AVX512-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; AVX512-NEXT: vmulps %ymm2, %ymm1, %ymm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
-; AVX512-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + ymm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm2
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm3
; AVX512-NEXT: retq
%div = fdiv fast <8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, %x
ret <8 x float> %div
@@ -849,8 +849,8 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; FMA-RECIP-LABEL: v8f32_one_step_2_divs:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - mem
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; FMA-RECIP-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1
; FMA-RECIP-NEXT: vmulps %ymm0, %ymm1, %ymm0
; FMA-RECIP-NEXT: retq
@@ -858,8 +858,8 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; BDVER2-LABEL: v8f32_one_step_2_divs:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1
-; BDVER2-NEXT: vfnmaddps {{.*}}(%rip), %ymm1, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm1, %ymm0, %ymm1, %ymm0
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %ymm1, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm1, %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1
; BDVER2-NEXT: vmulps %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: retq
@@ -892,8 +892,8 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - ymm2
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; HASWELL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1
; HASWELL-NEXT: vmulps %ymm0, %ymm1, %ymm0
; HASWELL-NEXT: retq
@@ -914,8 +914,8 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2
-; KNL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; KNL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - ymm2
+; KNL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; KNL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1
; KNL-NEXT: vmulps %ymm0, %ymm1, %ymm0
; KNL-NEXT: retq
@@ -923,8 +923,8 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; SKX-LABEL: v8f32_one_step_2_divs:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %ymm0, %ymm1
-; SKX-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
-; SKX-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
+; SKX-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm1 * ymm0) - mem
+; SKX-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm1) + ymm1
; SKX-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1
; SKX-NEXT: vmulps %ymm0, %ymm1, %ymm0
; SKX-NEXT: retq
@@ -987,24 +987,24 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; FMA-RECIP-LABEL: v8f32_two_step2:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; FMA-RECIP-NEXT: vfmadd231ps {{.*#+}} ymm2 = (ymm0 * ymm1) + ymm2
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; FMA-RECIP-NEXT: vfmsub231ps {{.*#+}} ymm2 = (ymm0 * ymm1) - ymm2
; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm2 = -(ymm2 * ymm1) + ymm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; FMA-RECIP-NEXT: vmulps %ymm1, %ymm2, %ymm3
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm1
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm1
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v8f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1
-; BDVER2-NEXT: vfmaddps {{.*}}(%rip), %ymm1, %ymm0, %ymm2
+; BDVER2-NEXT: vfmsubps {{.*}}(%rip), %ymm1, %ymm0, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; BDVER2-NEXT: vfnmaddps %ymm1, %ymm2, %ymm1, %ymm1
; BDVER2-NEXT: vmulps %ymm4, %ymm1, %ymm3
-; BDVER2-NEXT: vfnmaddps %ymm4, %ymm3, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm3, %ymm0, %ymm1, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm4, %ymm3, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm3, %ymm0, %ymm1, %ymm0
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v8f32_two_step2:
@@ -1042,13 +1042,13 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; HASWELL-LABEL: v8f32_two_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1
-; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; HASWELL-NEXT: vfmadd231ps {{.*#+}} ymm2 = (ymm0 * ymm1) + ymm2
+; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; HASWELL-NEXT: vfmsub231ps {{.*#+}} ymm2 = (ymm0 * ymm1) - ymm2
; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm2 = -(ymm2 * ymm1) + ymm1
; HASWELL-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; HASWELL-NEXT: vmulps %ymm1, %ymm2, %ymm3
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm1
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm1
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v8f32_two_step2:
@@ -1070,13 +1070,13 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; AVX512-LABEL: v8f32_two_step2:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcpps %ymm0, %ymm1
-; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; AVX512-NEXT: vfmadd231ps {{.*#+}} ymm2 = (ymm0 * ymm1) + ymm2
+; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; AVX512-NEXT: vfmsub231ps {{.*#+}} ymm2 = (ymm0 * ymm1) - ymm2
; AVX512-NEXT: vfnmadd132ps {{.*#+}} ymm2 = -(ymm2 * ymm1) + ymm1
; AVX512-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; AVX512-NEXT: vmulps %ymm1, %ymm2, %ymm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm1
-; AVX512-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm3 * ymm0) - ymm1
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
; AVX512-NEXT: retq
%div = fdiv fast <8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, %x
ret <8 x float> %div
@@ -1179,13 +1179,13 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; FMA-RECIP-NEXT: vmulps %ymm3, %ymm2, %ymm4
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm4
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm3
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm4
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; FMA-RECIP-NEXT: vmulps %ymm3, %ymm2, %ymm4
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm1 = (ymm2 * ymm1) + ymm4
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm3
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm4
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v16f32_one_step2:
@@ -1194,12 +1194,12 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; BDVER2-NEXT: vrcpps %ymm1, %ymm5
; BDVER2-NEXT: vmulps %ymm3, %ymm2, %ymm4
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm4, %ymm0, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm4, %ymm0, %ymm0
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
-; BDVER2-NEXT: vfmaddps %ymm4, %ymm0, %ymm2, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm4, %ymm0, %ymm2, %ymm0
; BDVER2-NEXT: vmulps %ymm3, %ymm5, %ymm4
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm4, %ymm1, %ymm1
-; BDVER2-NEXT: vfmaddps %ymm4, %ymm1, %ymm5, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm4, %ymm1, %ymm1
+; BDVER2-NEXT: vfnmaddps %ymm4, %ymm1, %ymm5, %ymm1
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v16f32_one_step2:
@@ -1243,13 +1243,13 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; HASWELL-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; HASWELL-NEXT: vmulps %ymm3, %ymm2, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm4 * ymm0) - ymm3
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm4
; HASWELL-NEXT: vrcpps %ymm1, %ymm2
; HASWELL-NEXT: vmovaps {{.*#+}} ymm3 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; HASWELL-NEXT: vmulps %ymm3, %ymm2, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm1 = (ymm2 * ymm1) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm3
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm4
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v16f32_one_step2:
@@ -1275,8 +1275,8 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; AVX512-NEXT: vrcp14ps %zmm0, %zmm1
; AVX512-NEXT: vmovaps {{.*#+}} zmm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0,9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; AVX512-NEXT: vmulps %zmm2, %zmm1, %zmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm3 * zmm0) + zmm2
-; AVX512-NEXT: vfmadd213ps {{.*#+}} zmm0 = (zmm1 * zmm0) + zmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm3 * zmm0) - zmm2
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm1 * zmm0) + zmm3
; AVX512-NEXT: retq
%div = fdiv fast <16 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0, float 9.0, float 10.0, float 11.0, float 12.0, float 13.0, float 14.0, float 15.0, float 16.0>, %x
ret <16 x float> %div
@@ -1346,11 +1346,11 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm2) + ymm2
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
-; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm3
+; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm2) + ymm2
; FMA-RECIP-NEXT: vmulps {{.*}}(%rip), %ymm1, %ymm2
; FMA-RECIP-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm3
; FMA-RECIP-NEXT: vmulps %ymm0, %ymm3, %ymm0
@@ -1361,12 +1361,12 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm0, %ymm2, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm0, %ymm2, %ymm0
; BDVER2-NEXT: vrcpps %ymm1, %ymm2
-; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm1, %ymm1
; BDVER2-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm3
-; BDVER2-NEXT: vfmaddps %ymm2, %ymm1, %ymm2, %ymm1
+; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm2, %ymm1
; BDVER2-NEXT: vmulps {{.*}}(%rip), %ymm1, %ymm2
; BDVER2-NEXT: vmulps %ymm0, %ymm3, %ymm0
; BDVER2-NEXT: vmulps %ymm1, %ymm2, %ymm1
@@ -1414,11 +1414,11 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm2 * ymm0) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm0 = -(ymm0 * ymm2) + ymm2
; HASWELL-NEXT: vrcpps %ymm1, %ymm2
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
-; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm2 * ymm1) - ymm3
+; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm1 = -(ymm1 * ymm2) + ymm2
; HASWELL-NEXT: vmulps {{.*}}(%rip), %ymm1, %ymm2
; HASWELL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm3
; HASWELL-NEXT: vmulps %ymm0, %ymm3, %ymm0
@@ -1447,8 +1447,8 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; AVX512-LABEL: v16f32_one_step_2_divs:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcp14ps %zmm0, %zmm1
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm1 * zmm0) + mem
-; AVX512-NEXT: vfmadd132ps {{.*#+}} zmm0 = (zmm0 * zmm1) + zmm1
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm1 * zmm0) - mem
+; AVX512-NEXT: vfnmadd132ps {{.*#+}} zmm0 = -(zmm0 * zmm1) + zmm1
; AVX512-NEXT: vmulps {{.*}}(%rip), %zmm0, %zmm1
; AVX512-NEXT: vmulps %zmm0, %zmm1, %zmm0
; AVX512-NEXT: retq
@@ -1552,40 +1552,40 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; FMA-RECIP-LABEL: v16f32_two_step2:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
+; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-RECIP-NEXT: vmovaps %ymm2, %ymm4
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm4 = (ymm0 * ymm4) + ymm3
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm0 * ymm4) - ymm3
; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; FMA-RECIP-NEXT: vmulps %ymm2, %ymm4, %ymm5
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm5 * ymm0) + ymm2
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm5
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm5 * ymm0) - ymm2
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm5
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
-; FMA-RECIP-NEXT: vfmadd231ps {{.*#+}} ymm3 = (ymm1 * ymm2) + ymm3
+; FMA-RECIP-NEXT: vfmsub231ps {{.*#+}} ymm3 = (ymm1 * ymm2) - ymm3
; FMA-RECIP-NEXT: vfnmadd132ps {{.*#+}} ymm3 = -(ymm3 * ymm2) + ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; FMA-RECIP-NEXT: vmulps %ymm2, %ymm3, %ymm4
-; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm2
-; FMA-RECIP-NEXT: vfmadd213ps {{.*#+}} ymm1 = (ymm3 * ymm1) + ymm4
+; FMA-RECIP-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm2
+; FMA-RECIP-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm3 * ymm1) + ymm4
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v16f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2
-; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; BDVER2-NEXT: vfmaddps %ymm3, %ymm2, %ymm0, %ymm4
+; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm0, %ymm4
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm4, %ymm2, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; BDVER2-NEXT: vmulps %ymm4, %ymm2, %ymm5
-; BDVER2-NEXT: vfnmaddps %ymm4, %ymm5, %ymm0, %ymm0
-; BDVER2-NEXT: vfmaddps %ymm5, %ymm0, %ymm2, %ymm0
+; BDVER2-NEXT: vfmsubps %ymm4, %ymm5, %ymm0, %ymm0
+; BDVER2-NEXT: vfnmaddps %ymm5, %ymm0, %ymm2, %ymm0
; BDVER2-NEXT: vrcpps %ymm1, %ymm2
; BDVER2-NEXT: vmovaps {{.*#+}} ymm5 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
-; BDVER2-NEXT: vfmaddps %ymm3, %ymm2, %ymm1, %ymm3
+; BDVER2-NEXT: vfmsubps %ymm3, %ymm2, %ymm1, %ymm3
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm3, %ymm2, %ymm2
; BDVER2-NEXT: vmulps %ymm5, %ymm2, %ymm4
-; BDVER2-NEXT: vfnmaddps %ymm5, %ymm4, %ymm1, %ymm1
-; BDVER2-NEXT: vfmaddps %ymm4, %ymm1, %ymm2, %ymm1
+; BDVER2-NEXT: vfmsubps %ymm5, %ymm4, %ymm1, %ymm1
+; BDVER2-NEXT: vfnmaddps %ymm4, %ymm1, %ymm2, %ymm1
; BDVER2-NEXT: retq
;
; BTVER2-LABEL: v16f32_two_step2:
@@ -1645,21 +1645,21 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; HASWELL-LABEL: v16f32_two_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm2
-; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
+; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NEXT: vmovaps %ymm2, %ymm4
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm4 = (ymm0 * ymm4) + ymm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm4 = (ymm0 * ymm4) - ymm3
; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm4 = -(ymm4 * ymm2) + ymm2
; HASWELL-NEXT: vmovaps {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; HASWELL-NEXT: vmulps %ymm2, %ymm4, %ymm5
; HASWELL-NEXT: vrcpps %ymm1, %ymm6
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm5 * ymm0) + ymm2
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm5
-; HASWELL-NEXT: vfmadd231ps {{.*#+}} ymm3 = (ymm1 * ymm6) + ymm3
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm0 = (ymm5 * ymm0) - ymm2
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm5
+; HASWELL-NEXT: vfmsub231ps {{.*#+}} ymm3 = (ymm1 * ymm6) - ymm3
; HASWELL-NEXT: vfnmadd132ps {{.*#+}} ymm3 = -(ymm3 * ymm6) + ymm6
; HASWELL-NEXT: vmovaps {{.*#+}} ymm2 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; HASWELL-NEXT: vmulps %ymm2, %ymm3, %ymm4
-; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm2
-; HASWELL-NEXT: vfmadd213ps {{.*#+}} ymm1 = (ymm3 * ymm1) + ymm4
+; HASWELL-NEXT: vfmsub213ps {{.*#+}} ymm1 = (ymm4 * ymm1) - ymm2
+; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm3 * ymm1) + ymm4
; HASWELL-NEXT: retq
;
; HASWELL-NO-FMA-LABEL: v16f32_two_step2:
@@ -1692,13 +1692,13 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; AVX512-LABEL: v16f32_two_step2:
; AVX512: # %bb.0:
; AVX512-NEXT: vrcp14ps %zmm0, %zmm1
-; AVX512-NEXT: vbroadcastss {{.*#+}} zmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
-; AVX512-NEXT: vfmadd231ps {{.*#+}} zmm2 = (zmm0 * zmm1) + zmm2
+; AVX512-NEXT: vbroadcastss {{.*#+}} zmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
+; AVX512-NEXT: vfmsub231ps {{.*#+}} zmm2 = (zmm0 * zmm1) - zmm2
; AVX512-NEXT: vfnmadd132ps {{.*#+}} zmm2 = -(zmm2 * zmm1) + zmm1
; AVX512-NEXT: vmovaps {{.*#+}} zmm1 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0,9.0E+0,1.0E+1,1.1E+1,1.2E+1,1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; AVX512-NEXT: vmulps %zmm1, %zmm2, %zmm3
-; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm3 * zmm0) + zmm1
-; AVX512-NEXT: vfmadd213ps {{.*#+}} zmm0 = (zmm2 * zmm0) + zmm3
+; AVX512-NEXT: vfmsub213ps {{.*#+}} zmm0 = (zmm3 * zmm0) - zmm1
+; AVX512-NEXT: vfnmadd213ps {{.*#+}} zmm0 = -(zmm2 * zmm0) + zmm3
; AVX512-NEXT: retq
%div = fdiv fast <16 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0, float 9.0, float 10.0, float 11.0, float 12.0, float 13.0, float 14.0, float 15.0, float 16.0>, %x
ret <16 x float> %div
More information about the llvm-commits
mailing list