RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-01-31 12:41:49 +01:00
Code Issues Projects Releases Wiki Activity

[X86] Force floating point values in constant pool decoding to print in scientific notation so they can't be confused with integers.

Browse Source 
When the floating point constants are whole numbers they have no decimal point so look like integers, but mean something very different in something like an 'and' instruction.

Ideally we would just print a decimal point and a 0, but I couldn't see how to make APFloat::toString do that.

llvm-svn: 345488
This commit is contained in:
Craig Topper 2018-10-29 04:52:04 +00:00
parent 0fddf00499
commit 365c4a3fad
49 changed files with 492 additions and 491 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
lib/Target/X86/X86MCInstLower.cpp
View File

@ -1499,7 +1499,8 @@ static void printConstant(const APInt &Val, raw_ostream &CS) {
static void printConstant(const APFloat &Flt, raw_ostream &CS) {
SmallString<32> Str;
Flt.toString(Str);
// Force scientific notation to distinquish from integers.
Flt.toString(Str, 0, 0);
CS << Str;
}

4
test/CodeGen/X86/2011-10-19-widen_vselect.ll
View File

@ -72,7 +72,7 @@ define void @full_test() {
; X32-NEXT: cvtdq2ps %xmm0, %xmm1
; X32-NEXT: xorps %xmm0, %xmm0
; X32-NEXT: cmpltps %xmm2, %xmm0
; X32-NEXT: movaps {{.*#+}} xmm3 = <1,1,u,u>
; X32-NEXT: movaps {{.*#+}} xmm3 = <1.0E+0,1.0E+0,u,u>
; X32-NEXT: addps %xmm1, %xmm3
; X32-NEXT: movaps %xmm1, %xmm4
; X32-NEXT: blendvps %xmm0, %xmm3, %xmm4
@ -92,7 +92,7 @@ define void @full_test() {
; X64-NEXT: cvtdq2ps %xmm0, %xmm1
; X64-NEXT: xorps %xmm0, %xmm0
; X64-NEXT: cmpltps %xmm2, %xmm0
; X64-NEXT: movaps {{.*#+}} xmm3 = <1,1,u,u>
; X64-NEXT: movaps {{.*#+}} xmm3 = <1.0E+0,1.0E+0,u,u>
; X64-NEXT: addps %xmm1, %xmm3
; X64-NEXT: movaps %xmm1, %xmm4
; X64-NEXT: blendvps %xmm0, %xmm3, %xmm4

2
test/CodeGen/X86/2011-20-21-zext-ui2fp.ll
View File

@ -7,7 +7,7 @@
define void @ui_to_fp_conv(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) nounwind {
; CHECK-LABEL: ui_to_fp_conv:
; CHECK: # %bb.0: # %allocas
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1,1,0,0]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.0E+0,0.0E+0,0.0E+0]
; CHECK-NEXT: xorps %xmm1, %xmm1
; CHECK-NEXT: movups %xmm1, 16(%rsi)
; CHECK-NEXT: movups %xmm0, (%rsi)

2
test/CodeGen/X86/avx-basic.ll
View File

@ -92,7 +92,7 @@ define <8 x i32> @VMOVZQI2PQI([0 x float]* nocapture %aFOO) nounwind {
define <16 x float> @fneg(<16 x float> %a) nounwind {
; CHECK-LABEL: fneg:
; CHECK: ## %bb.0:
; CHECK-NEXT: vmovaps {{.*#+}} ymm2 = [-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vmovaps {{.*#+}} ymm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vxorps %ymm2, %ymm0, %ymm0
; CHECK-NEXT: vxorps %ymm2, %ymm1, %ymm1
; CHECK-NEXT: retq

4
test/CodeGen/X86/avx-vbroadcast.ll
View File

@ -316,12 +316,12 @@ entry:
define <4 x float> @_e2(float* %ptr) nounwind uwtable readnone ssp {
; X32-LABEL: _e2:
; X32: ## %bb.0: ## %entry
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X32-NEXT: retl
;
; X64-LABEL: _e2:
; X64: ## %bb.0: ## %entry
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X64-NEXT: retq
entry:
%vecinit.i = insertelement <4 x float> undef, float 0xbf80000000000000, i32 0

8
test/CodeGen/X86/avx-vperm2x128.ll
View File

@ -532,7 +532,7 @@ define <4 x double> @ld0_hi0_lo1_4f64(<4 x double> * %pa, <4 x double> %b) nounw
; AVX2-LABEL: ld0_hi0_lo1_4f64:
; AVX2: # %bb.0: # %entry
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1,1,1,1]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX2-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
@ -552,7 +552,7 @@ define <4 x double> @ld1_hi0_hi1_4f64(<4 x double> %a, <4 x double> * %pb) nounw
; AVX2-LABEL: ld1_hi0_hi1_4f64:
; AVX2: # %bb.0: # %entry
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1,1,1,1]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX2-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
@ -572,7 +572,7 @@ define <8 x float> @ld0_hi0_lo1_8f32(<8 x float> * %pa, <8 x float> %b) nounwind
; AVX2-LABEL: ld0_hi0_lo1_8f32:
; AVX2: # %bb.0: # %entry
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = mem[2,3],ymm0[0,1]
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:
@ -592,7 +592,7 @@ define <8 x float> @ld1_hi0_hi1_8f32(<8 x float> %a, <8 x float> * %pb) nounwind
; AVX2-LABEL: ld1_hi0_hi1_8f32:
; AVX2: # %bb.0: # %entry
; AVX2-NEXT: vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],mem[2,3]
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX2-NEXT: retq
entry:

4
test/CodeGen/X86/avx2-fma-fneg-combine.ll
View File

@ -44,14 +44,14 @@ define <4 x float> @test3(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
; X32-LABEL: test3:
; X32: # %bb.0: # %entry
; X32-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
; X32-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; X32-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; X32-NEXT: vxorps %xmm1, %xmm0, %xmm0
; X32-NEXT: retl
;
; X64-LABEL: test3:
; X64: # %bb.0: # %entry
; X64-NEXT: vfnmadd213ss {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2
; X64-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; X64-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; X64-NEXT: vxorps %xmm1, %xmm0, %xmm0
; X64-NEXT: retq
entry:

8
test/CodeGen/X86/avx2-vbroadcast.ll
View File

@ -615,13 +615,13 @@ entry:
define <8 x float> @V113(<8 x float> %in) nounwind uwtable readnone ssp {
; X32-AVX2-LABEL: V113:
; X32-AVX2: ## %bb.0: ## %entry
; X32-AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X32-AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X32-AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; X32-AVX2-NEXT: retl
;
; X64-AVX2-LABEL: V113:
; X64-AVX2: ## %bb.0: ## %entry
; X64-AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X64-AVX2-NEXT: vbroadcastss {{.*#+}} ymm1 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X64-AVX2-NEXT: vaddps %ymm1, %ymm0, %ymm0
; X64-AVX2-NEXT: retq
;
@ -642,12 +642,12 @@ entry:
define <4 x float> @_e2(float* %ptr) nounwind uwtable readnone ssp {
; X32-LABEL: _e2:
; X32: ## %bb.0:
; X32-NEXT: vbroadcastss {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X32-NEXT: vbroadcastss {{.*#+}} xmm0 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X32-NEXT: retl
;
; X64-LABEL: _e2:
; X64: ## %bb.0:
; X64-NEXT: vbroadcastss {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
; X64-NEXT: vbroadcastss {{.*#+}} xmm0 = [-7.8125E-3,-7.8125E-3,-7.8125E-3,-7.8125E-3]
; X64-NEXT: retq
%vecinit.i = insertelement <4 x float> undef, float 0xbf80000000000000, i32 0
%vecinit2.i = insertelement <4 x float> %vecinit.i, float 0xbf80000000000000, i32 1

6
test/CodeGen/X86/avx512-arith.ll
View File

@ -969,7 +969,7 @@ define <16 x float> @test_fxor(<16 x float> %a) {
define <8 x float> @test_fxor_8f32(<8 x float> %a) {
; AVX512F-LABEL: test_fxor_8f32:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0,-0,-0,-0,-0,-0,-0,-0]
; AVX512F-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX512F-NEXT: vxorps %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: retq
;
@ -980,13 +980,13 @@ define <8 x float> @test_fxor_8f32(<8 x float> %a) {
;
; AVX512BW-LABEL: test_fxor_8f32:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0,-0,-0,-0,-0,-0,-0,-0]
; AVX512BW-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX512BW-NEXT: vxorps %ymm1, %ymm0, %ymm0
; AVX512BW-NEXT: retq
;
; AVX512DQ-LABEL: test_fxor_8f32:
; AVX512DQ: # %bb.0:
; AVX512DQ-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0,-0,-0,-0,-0,-0,-0,-0]
; AVX512DQ-NEXT: vbroadcastss {{.*#+}} ymm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX512DQ-NEXT: vxorps %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: retq
;

16
test/CodeGen/X86/avx512-intrinsics-fast-isel.ll
View File

@ -3156,7 +3156,7 @@ entry:
define <8 x double> @test_mm512_fnmsub_round_pd(<8 x double> %__A, <8 x double> %__B, <8 x double> %__C) {
; CHECK-LABEL: test_mm512_fnmsub_round_pd:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vpbroadcastq {{.*#+}} zmm3 = [-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vpbroadcastq {{.*#+}} zmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vpxorq %zmm3, %zmm0, %zmm4
; CHECK-NEXT: vpxorq %zmm3, %zmm2, %zmm0
; CHECK-NEXT: vfmadd231pd {rn-sae}, %zmm4, %zmm1, %zmm0
@ -3387,7 +3387,7 @@ entry:
define <8 x double> @test_mm512_fnmsub_pd(<8 x double> %__A, <8 x double> %__B, <8 x double> %__C) {
; CHECK-LABEL: test_mm512_fnmsub_pd:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vpbroadcastq {{.*#+}} zmm3 = [-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vpbroadcastq {{.*#+}} zmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vpxorq %zmm3, %zmm0, %zmm4
; CHECK-NEXT: vpxorq %zmm3, %zmm2, %zmm0
; CHECK-NEXT: vfmadd231pd {{.*#+}} zmm0 = (zmm1 * zmm4) + zmm0
@ -3613,7 +3613,7 @@ entry:
define <16 x float> @test_mm512_fnmsub_round_ps(<16 x float> %__A, <16 x float> %__B, <16 x float> %__C) {
; CHECK-LABEL: test_mm512_fnmsub_round_ps:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vpbroadcastd {{.*#+}} zmm3 = [-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vpbroadcastd {{.*#+}} zmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vpxord %zmm3, %zmm0, %zmm4
; CHECK-NEXT: vpxord %zmm3, %zmm2, %zmm0
; CHECK-NEXT: vfmadd231ps {rn-sae}, %zmm4, %zmm1, %zmm0
@ -3836,7 +3836,7 @@ entry:
define <16 x float> @test_mm512_fnmsub_ps(<16 x float> %__A, <16 x float> %__B, <16 x float> %__C) {
; CHECK-LABEL: test_mm512_fnmsub_ps:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vpbroadcastd {{.*#+}} zmm3 = [-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vpbroadcastd {{.*#+}} zmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vpxord %zmm3, %zmm0, %zmm4
; CHECK-NEXT: vpxord %zmm3, %zmm2, %zmm0
; CHECK-NEXT: vfmadd231ps {{.*#+}} zmm0 = (zmm1 * zmm4) + zmm0
@ -7459,7 +7459,7 @@ define double @test_mm512_mask_reduce_mul_pd(i8 zeroext %__M, <8 x double> %__W)
; X86-NEXT: subl $8, %esp
; X86-NEXT: movb 8(%ebp), %al
; X86-NEXT: kmovw %eax, %k1
; X86-NEXT: vbroadcastsd {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1]
; X86-NEXT: vbroadcastsd {{.*#+}} zmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; X86-NEXT: vmovapd %zmm0, %zmm1 {%k1}
; X86-NEXT: vextractf64x4 $1, %zmm1, %ymm0
; X86-NEXT: vmulpd %ymm0, %ymm1, %ymm0
@ -7478,7 +7478,7 @@ define double @test_mm512_mask_reduce_mul_pd(i8 zeroext %__M, <8 x double> %__W)
; X64-LABEL: test_mm512_mask_reduce_mul_pd:
; X64: # %bb.0: # %entry
; X64-NEXT: kmovw %edi, %k1
; X64-NEXT: vbroadcastsd {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1]
; X64-NEXT: vbroadcastsd {{.*#+}} zmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; X64-NEXT: vmovapd %zmm0, %zmm1 {%k1}
; X64-NEXT: vextractf64x4 $1, %zmm1, %ymm0
; X64-NEXT: vmulpd %ymm0, %ymm1, %ymm0
@ -7565,7 +7565,7 @@ define float @test_mm512_mask_reduce_mul_ps(i16 zeroext %__M, <16 x float> %__W)
; X86-NEXT: pushl %eax
; X86-NEXT: .cfi_def_cfa_offset 8
; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1
; X86-NEXT: vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; X86-NEXT: vbroadcastss {{.*#+}} zmm1 = [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]
; X86-NEXT: vmovaps %zmm0, %zmm1 {%k1}
; X86-NEXT: vextractf64x4 $1, %zmm1, %ymm0
; X86-NEXT: vmulps %ymm0, %ymm1, %ymm0
@ -7585,7 +7585,7 @@ define float @test_mm512_mask_reduce_mul_ps(i16 zeroext %__M, <16 x float> %__W)
; X64-LABEL: test_mm512_mask_reduce_mul_ps:
; X64: # %bb.0: # %entry
; X64-NEXT: kmovw %edi, %k1
; X64-NEXT: vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; X64-NEXT: vbroadcastss {{.*#+}} zmm1 = [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]
; X64-NEXT: vmovaps %zmm0, %zmm1 {%k1}
; X64-NEXT: vextractf64x4 $1, %zmm1, %ymm0
; X64-NEXT: vmulps %ymm0, %ymm1, %ymm0

84
test/CodeGen/X86/broadcast-elm-cross-splat-vec.ll
View File

@ -1600,7 +1600,7 @@ define <8 x i64> @f8xi64_i256(<8 x i64> %a) {
define <4 x float> @f4xf32_f64(<4 x float> %a) {
; AVX-LABEL: f4xf32_f64:
; AVX: # %bb.0:
; AVX-NEXT: vmovddup {{.*#+}} xmm1 = [0.0078125018626451492,0.0078125018626451492]
; AVX-NEXT: vmovddup {{.*#+}} xmm1 = [7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-NEXT: # xmm1 = mem[0,0]
; AVX-NEXT: vaddps %xmm1, %xmm0, %xmm0
; AVX-NEXT: vdivps %xmm0, %xmm1, %xmm0
@ -1608,7 +1608,7 @@ define <4 x float> @f4xf32_f64(<4 x float> %a) {
;
; ALL32-LABEL: f4xf32_f64:
; ALL32: # %bb.0:
; ALL32-NEXT: vmovddup {{.*#+}} xmm1 = [0.0078125018626451492,0.0078125018626451492]
; ALL32-NEXT: vmovddup {{.*#+}} xmm1 = [7.8125018626451492E-3,7.8125018626451492E-3]
; ALL32-NEXT: # xmm1 = mem[0,0]
; ALL32-NEXT: vaddps %xmm1, %xmm0, %xmm0
; ALL32-NEXT: vdivps %xmm0, %xmm1, %xmm0
@ -1616,7 +1616,7 @@ define <4 x float> @f4xf32_f64(<4 x float> %a) {
;
; AVX-64-LABEL: f4xf32_f64:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vmovddup {{.*#+}} xmm1 = [0.0078125018626451492,0.0078125018626451492]
; AVX-64-NEXT: vmovddup {{.*#+}} xmm1 = [7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-64-NEXT: # xmm1 = mem[0,0]
; AVX-64-NEXT: vaddps %xmm1, %xmm0, %xmm0
; AVX-64-NEXT: vdivps %xmm0, %xmm1, %xmm0
@ -1637,21 +1637,21 @@ define <4 x float> @f4xf32_f64(<4 x float> %a) {
define <8 x float> @f8xf32_f64(<8 x float> %a) {
; AVX-LABEL: f8xf32_f64:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm1 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm1 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX-NEXT: vdivps %ymm0, %ymm1, %ymm0
; AVX-NEXT: retl
;
; ALL32-LABEL: f8xf32_f64:
; ALL32: # %bb.0:
; ALL32-NEXT: vbroadcastsd {{.*#+}} ymm1 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; ALL32-NEXT: vbroadcastsd {{.*#+}} ymm1 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; ALL32-NEXT: vaddps %ymm1, %ymm0, %ymm0
; ALL32-NEXT: vdivps %ymm0, %ymm1, %ymm0
; ALL32-NEXT: retl
;
; AVX-64-LABEL: f8xf32_f64:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastsd {{.*#+}} ymm1 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX-64-NEXT: vbroadcastsd {{.*#+}} ymm1 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-64-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX-64-NEXT: vdivps %ymm0, %ymm1, %ymm0
; AVX-64-NEXT: retq
@ -1671,7 +1671,7 @@ define <8 x float> @f8xf32_f64(<8 x float> %a) {
define <8 x float> @f8xf32_f128(<8 x float> %a) {
; AVX-LABEL: f8xf32_f128:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-NEXT: # ymm1 = mem[0,1,0,1]
; AVX-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX-NEXT: vdivps %ymm0, %ymm1, %ymm0
@ -1679,7 +1679,7 @@ define <8 x float> @f8xf32_f128(<8 x float> %a) {
;
; ALL32-LABEL: f8xf32_f128:
; ALL32: # %bb.0:
; ALL32-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
; ALL32-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; ALL32-NEXT: # ymm1 = mem[0,1,0,1]
; ALL32-NEXT: vaddps %ymm1, %ymm0, %ymm0
; ALL32-NEXT: vdivps %ymm0, %ymm1, %ymm0
@ -1687,7 +1687,7 @@ define <8 x float> @f8xf32_f128(<8 x float> %a) {
;
; AVX-64-LABEL: f8xf32_f128:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-64-NEXT: # ymm1 = mem[0,1,0,1]
; AVX-64-NEXT: vaddps %ymm1, %ymm0, %ymm0
; AVX-64-NEXT: vdivps %ymm0, %ymm1, %ymm0
@ -1695,7 +1695,7 @@ define <8 x float> @f8xf32_f128(<8 x float> %a) {
;
; ALL64-LABEL: f8xf32_f128:
; ALL64: # %bb.0:
; ALL64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
; ALL64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; ALL64-NEXT: # ymm1 = mem[0,1,0,1]
; ALL64-NEXT: vaddps %ymm1, %ymm0, %ymm0
; ALL64-NEXT: vdivps %ymm0, %ymm1, %ymm0
@ -1709,7 +1709,7 @@ define <8 x float> @f8xf32_f128(<8 x float> %a) {
define <16 x float> @f16xf32_f64(<16 x float> %a) {
; AVX-LABEL: f16xf32_f64:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1718,7 +1718,7 @@ define <16 x float> @f16xf32_f64(<16 x float> %a) {
;
; AVX2-LABEL: f16xf32_f64:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX2-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1727,14 +1727,14 @@ define <16 x float> @f16xf32_f64(<16 x float> %a) {
;
; AVX512-LABEL: f16xf32_f64:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcastsd {{.*#+}} zmm1 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX512-NEXT: vbroadcastsd {{.*#+}} zmm1 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX512-NEXT: vaddps %zmm1, %zmm0, %zmm0
; AVX512-NEXT: vdivps %zmm0, %zmm1, %zmm0
; AVX512-NEXT: retl
;
; AVX-64-LABEL: f16xf32_f64:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastsd {{.*#+}} ymm2 = [0.0078125018626451492,0.0078125018626451492,0.0078125018626451492,0.0078125018626451492]
; AVX-64-NEXT: vbroadcastsd {{.*#+}} ymm2 = [7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3,7.8125018626451492E-3]
; AVX-64-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-64-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX-64-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1765,7 +1765,7 @@ define <16 x float> @f16xf32_f64(<16 x float> %a) {
define <16 x float> @f16xf32_f128(<16 x float> %a) {
; AVX-LABEL: f16xf32_f128:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-NEXT: # ymm2 = mem[0,1,0,1]
; AVX-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-NEXT: vaddps %ymm2, %ymm0, %ymm0
@ -1775,7 +1775,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
;
; AVX2-LABEL: f16xf32_f128:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX2-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vaddps %ymm2, %ymm0, %ymm0
@ -1785,7 +1785,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
;
; AVX512-LABEL: f16xf32_f128:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3,4,1,2,3,4,1,2,3]
; AVX512-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX512-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; AVX512-NEXT: vaddps %zmm1, %zmm0, %zmm0
; AVX512-NEXT: vdivps %zmm0, %zmm1, %zmm0
@ -1793,7 +1793,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
;
; AVX-64-LABEL: f16xf32_f128:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-64-NEXT: # ymm2 = mem[0,1,0,1]
; AVX-64-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-64-NEXT: vaddps %ymm2, %ymm0, %ymm0
@ -1803,7 +1803,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
;
; AVX2-64-LABEL: f16xf32_f128:
; AVX2-64: # %bb.0:
; AVX2-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
; AVX2-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX2-64-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-64-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX2-64-NEXT: vaddps %ymm2, %ymm0, %ymm0
@ -1813,7 +1813,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
;
; AVX512F-64-LABEL: f16xf32_f128:
; AVX512F-64: # %bb.0:
; AVX512F-64-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3,4,1,2,3,4,1,2,3]
; AVX512F-64-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX512F-64-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; AVX512F-64-NEXT: vaddps %zmm1, %zmm0, %zmm0
; AVX512F-64-NEXT: vdivps %zmm0, %zmm1, %zmm0
@ -1827,7 +1827,7 @@ define <16 x float> @f16xf32_f128(<16 x float> %a) {
define <16 x float> @f16xf32_f256(<16 x float> %a) {
; AVX-LABEL: f16xf32_f256:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
; AVX-NEXT: vmovaps {{.*#+}} ymm2 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1836,7 +1836,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
;
; AVX2-LABEL: f16xf32_f256:
; AVX2: # %bb.0:
; AVX2-NEXT: vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
; AVX2-NEXT: vmovaps {{.*#+}} ymm2 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX2-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1845,7 +1845,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
;
; AVX512-LABEL: f16xf32_f256:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7]
; AVX512-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX512-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vaddps %zmm1, %zmm0, %zmm0
; AVX512-NEXT: vdivps %zmm0, %zmm1, %zmm0
@ -1853,7 +1853,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
;
; AVX-64-LABEL: f16xf32_f256:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
; AVX-64-NEXT: vmovaps {{.*#+}} ymm2 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX-64-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX-64-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX-64-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1862,7 +1862,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
;
; AVX2-64-LABEL: f16xf32_f256:
; AVX2-64: # %bb.0:
; AVX2-64-NEXT: vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
; AVX2-64-NEXT: vmovaps {{.*#+}} ymm2 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX2-64-NEXT: vaddps %ymm2, %ymm1, %ymm1
; AVX2-64-NEXT: vaddps %ymm2, %ymm0, %ymm0
; AVX2-64-NEXT: vdivps %ymm0, %ymm2, %ymm0
@ -1871,7 +1871,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
;
; AVX512F-64-LABEL: f16xf32_f256:
; AVX512F-64: # %bb.0:
; AVX512F-64-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7]
; AVX512F-64-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0,8.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; AVX512F-64-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3]
; AVX512F-64-NEXT: vaddps %zmm1, %zmm0, %zmm0
; AVX512F-64-NEXT: vdivps %zmm0, %zmm1, %zmm0
@ -1885,7 +1885,7 @@ define <16 x float> @f16xf32_f256(<16 x float> %a) {
define <4 x double> @f4xf64_f128(<4 x double> %a) {
; AVX-LABEL: f4xf64_f128:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX-NEXT: # ymm1 = mem[0,1,0,1]
; AVX-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX-NEXT: vdivpd %ymm0, %ymm1, %ymm0
@ -1893,7 +1893,7 @@ define <4 x double> @f4xf64_f128(<4 x double> %a) {
;
; ALL32-LABEL: f4xf64_f128:
; ALL32: # %bb.0:
; ALL32-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
; ALL32-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; ALL32-NEXT: # ymm1 = mem[0,1,0,1]
; ALL32-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; ALL32-NEXT: vdivpd %ymm0, %ymm1, %ymm0
@ -1901,7 +1901,7 @@ define <4 x double> @f4xf64_f128(<4 x double> %a) {
;
; AVX-64-LABEL: f4xf64_f128:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX-64-NEXT: # ymm1 = mem[0,1,0,1]
; AVX-64-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; AVX-64-NEXT: vdivpd %ymm0, %ymm1, %ymm0
@ -1909,7 +1909,7 @@ define <4 x double> @f4xf64_f128(<4 x double> %a) {
;
; ALL64-LABEL: f4xf64_f128:
; ALL64: # %bb.0:
; ALL64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
; ALL64-NEXT: vbroadcastf128 {{.*#+}} ymm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; ALL64-NEXT: # ymm1 = mem[0,1,0,1]
; ALL64-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; ALL64-NEXT: vdivpd %ymm0, %ymm1, %ymm0
@ -1923,7 +1923,7 @@ define <4 x double> @f4xf64_f128(<4 x double> %a) {
define <8 x double> @f8xf64_f128(<8 x double> %a) {
; AVX-LABEL: f8xf64_f128:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
; AVX-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX-NEXT: # ymm2 = mem[0,1,0,1]
; AVX-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX-NEXT: vaddpd %ymm2, %ymm0, %ymm0
@ -1933,7 +1933,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
;
; AVX2-LABEL: f8xf64_f128:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
; AVX2-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX2-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vaddpd %ymm2, %ymm0, %ymm0
@ -1943,7 +1943,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
;
; AVX512-LABEL: f8xf64_f128:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [2,1,2,1,2,1,2,1]
; AVX512-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0,2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX512-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; AVX512-NEXT: vaddpd %zmm1, %zmm0, %zmm0
; AVX512-NEXT: vdivpd %zmm0, %zmm1, %zmm0
@ -1951,7 +1951,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
;
; AVX-64-LABEL: f8xf64_f128:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
; AVX-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX-64-NEXT: # ymm2 = mem[0,1,0,1]
; AVX-64-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX-64-NEXT: vaddpd %ymm2, %ymm0, %ymm0
@ -1961,7 +1961,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
;
; AVX2-64-LABEL: f8xf64_f128:
; AVX2-64: # %bb.0:
; AVX2-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
; AVX2-64-NEXT: vbroadcastf128 {{.*#+}} ymm2 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX2-64-NEXT: # ymm2 = mem[0,1,0,1]
; AVX2-64-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX2-64-NEXT: vaddpd %ymm2, %ymm0, %ymm0
@ -1971,7 +1971,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
;
; AVX512F-64-LABEL: f8xf64_f128:
; AVX512F-64: # %bb.0:
; AVX512F-64-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [2,1,2,1,2,1,2,1]
; AVX512F-64-NEXT: vbroadcastf32x4 {{.*#+}} zmm1 = [2.0E+0,1.0E+0,2.0E+0,1.0E+0,2.0E+0,1.0E+0,2.0E+0,1.0E+0]
; AVX512F-64-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
; AVX512F-64-NEXT: vaddpd %zmm1, %zmm0, %zmm0
; AVX512F-64-NEXT: vdivpd %zmm0, %zmm1, %zmm0
@ -1992,7 +1992,7 @@ define <8 x double> @f8xf64_f128(<8 x double> %a) {
define <8 x double> @f8xf64_f256(<8 x double> %a) {
; AVX-LABEL: f8xf64_f256:
; AVX: # %bb.0:
; AVX-NEXT: vmovapd {{.*#+}} ymm2 = [4,1,2,3]
; AVX-NEXT: vmovapd {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; AVX-NEXT: vdivpd %ymm0, %ymm2, %ymm0
@ -2001,7 +2001,7 @@ define <8 x double> @f8xf64_f256(<8 x double> %a) {
;
; AVX2-LABEL: f8xf64_f256:
; AVX2: # %bb.0:
; AVX2-NEXT: vmovapd {{.*#+}} ymm2 = [4,1,2,3]
; AVX2-NEXT: vmovapd {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX2-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; AVX2-NEXT: vdivpd %ymm0, %ymm2, %ymm0
@ -2010,7 +2010,7 @@ define <8 x double> @f8xf64_f256(<8 x double> %a) {
;
; AVX512-LABEL: f8xf64_f256:
; AVX512: # %bb.0:
; AVX512-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3]
; AVX512-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX512-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3]
; AVX512-NEXT: vaddpd %zmm1, %zmm0, %zmm0
; AVX512-NEXT: vdivpd %zmm0, %zmm1, %zmm0
@ -2018,7 +2018,7 @@ define <8 x double> @f8xf64_f256(<8 x double> %a) {
;
; AVX-64-LABEL: f8xf64_f256:
; AVX-64: # %bb.0:
; AVX-64-NEXT: vmovapd {{.*#+}} ymm2 = [4,1,2,3]
; AVX-64-NEXT: vmovapd {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX-64-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX-64-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; AVX-64-NEXT: vdivpd %ymm0, %ymm2, %ymm0
@ -2027,7 +2027,7 @@ define <8 x double> @f8xf64_f256(<8 x double> %a) {
;
; AVX2-64-LABEL: f8xf64_f256:
; AVX2-64: # %bb.0:
; AVX2-64-NEXT: vmovapd {{.*#+}} ymm2 = [4,1,2,3]
; AVX2-64-NEXT: vmovapd {{.*#+}} ymm2 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX2-64-NEXT: vaddpd %ymm2, %ymm1, %ymm1
; AVX2-64-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; AVX2-64-NEXT: vdivpd %ymm0, %ymm2, %ymm0
@ -2036,7 +2036,7 @@ define <8 x double> @f8xf64_f256(<8 x double> %a) {
;
; AVX512F-64-LABEL: f8xf64_f256:
; AVX512F-64: # %bb.0:
; AVX512F-64-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3]
; AVX512F-64-NEXT: vbroadcastf64x4 {{.*#+}} zmm1 = [4.0E+0,1.0E+0,2.0E+0,3.0E+0,4.0E+0,1.0E+0,2.0E+0,3.0E+0]
; AVX512F-64-NEXT: # zmm1 = mem[0,1,2,3,0,1,2,3]
; AVX512F-64-NEXT: vaddpd %zmm1, %zmm0, %zmm0
; AVX512F-64-NEXT: vdivpd %zmm0, %zmm1, %zmm0

2
test/CodeGen/X86/buildvec-insertvec.ll
View File

@ -65,7 +65,7 @@ entry:
define <2 x double> @test_negative_zero_2(<2 x double> %A) {
; SSE2-LABEL: test_negative_zero_2:
; SSE2: # %bb.0: # %entry
; SSE2-NEXT: movapd {{.*#+}} xmm1 = <u,-0>
; SSE2-NEXT: movapd {{.*#+}} xmm1 = <u,-0.0E+0>
; SSE2-NEXT: movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE2-NEXT: movapd %xmm1, %xmm0
; SSE2-NEXT: retq

4
test/CodeGen/X86/combine-fabs.ll
View File

@ -24,12 +24,12 @@ define float @combine_fabs_constant() {
define <4 x float> @combine_vec_fabs_constant() {
; SSE-LABEL: combine_vec_fabs_constant:
; SSE: # %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm0 = [0,0,2,2]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [0.0E+0,0.0E+0,2.0E+0,2.0E+0]
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_fabs_constant:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,0,2,2]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0.0E+0,0.0E+0,2.0E+0,2.0E+0]
; AVX-NEXT: retq
%1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> <float 0.0, float -0.0, float 2.0, float -2.0>)
ret <4 x float> %1

22
test/CodeGen/X86/combine-fcopysign.ll
View File

@ -62,7 +62,7 @@ define <4 x float> @combine_vec_fcopysign_neg_constant0(<4 x float> %x) {
;
; AVX-LABEL: combine_vec_fcopysign_neg_constant0:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vorps %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> <float -2.0, float -2.0, float -2.0, float -2.0>)
@ -77,7 +77,7 @@ define <4 x float> @combine_vec_fcopysign_neg_constant1(<4 x float> %x) {
;
; AVX-LABEL: combine_vec_fcopysign_neg_constant1:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vorps %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> <float -0.0, float -2.0, float -4.0, float -8.0>)
@ -92,7 +92,7 @@ define <4 x float> @combine_vec_fcopysign_fneg_fabs_sgn(<4 x float> %x, <4 x flo
;
; AVX-LABEL: combine_vec_fcopysign_fneg_fabs_sgn:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vorps %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %y)
@ -112,7 +112,7 @@ define <4 x float> @combine_vec_fcopysign_fabs_mag(<4 x float> %x, <4 x float> %
;
; AVX-LABEL: combine_vec_fcopysign_fabs_mag:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandps %xmm2, %xmm1, %xmm1
; AVX-NEXT: vbroadcastss {{.*}}(%rip), %xmm2
; AVX-NEXT: vandps %xmm2, %xmm0, %xmm0
@ -134,7 +134,7 @@ define <4 x float> @combine_vec_fcopysign_fneg_mag(<4 x float> %x, <4 x float> %
;
; AVX-LABEL: combine_vec_fcopysign_fneg_mag:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandps %xmm2, %xmm1, %xmm1
; AVX-NEXT: vbroadcastss {{.*}}(%rip), %xmm2
; AVX-NEXT: vandps %xmm2, %xmm0, %xmm0
@ -156,7 +156,7 @@ define <4 x float> @combine_vec_fcopysign_fcopysign_mag(<4 x float> %x, <4 x flo
;
; AVX-LABEL: combine_vec_fcopysign_fcopysign_mag:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandps %xmm2, %xmm1, %xmm1
; AVX-NEXT: vbroadcastss {{.*}}(%rip), %xmm2
; AVX-NEXT: vandps %xmm2, %xmm0, %xmm0
@ -178,7 +178,7 @@ define <4 x float> @combine_vec_fcopysign_fcopysign_sgn(<4 x float> %x, <4 x flo
;
; AVX-LABEL: combine_vec_fcopysign_fcopysign_sgn:
; AVX: # %bb.0:
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandps %xmm1, %xmm2, %xmm1
; AVX-NEXT: vbroadcastss {{.*}}(%rip), %xmm2
; AVX-NEXT: vandps %xmm2, %xmm0, %xmm0
@ -202,7 +202,7 @@ define <4 x double> @combine_vec_fcopysign_fpext_sgn(<4 x double> %x, <4 x float
; SSE-NEXT: movaps {{.*#+}} xmm7
; SSE-NEXT: movaps %xmm0, %xmm2
; SSE-NEXT: andps %xmm7, %xmm2
; SSE-NEXT: movaps {{.*#+}} xmm8 = [-0,-0]
; SSE-NEXT: movaps {{.*#+}} xmm8 = [-0.0E+0,-0.0E+0]
; SSE-NEXT: andps %xmm8, %xmm4
; SSE-NEXT: orps %xmm4, %xmm2
; SSE-NEXT: movhlps {{.*#+}} xmm0 = xmm0[1,1]
@ -232,7 +232,7 @@ define <4 x double> @combine_vec_fcopysign_fpext_sgn(<4 x double> %x, <4 x float
; AVX-NEXT: vbroadcastsd {{.*}}(%rip), %ymm2
; AVX-NEXT: vandps %ymm2, %ymm0, %ymm0
; AVX-NEXT: vcvtps2pd %xmm1, %ymm1
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastsd {{.*#+}} ymm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandps %ymm2, %ymm1, %ymm1
; AVX-NEXT: vorps %ymm1, %ymm0, %ymm0
; AVX-NEXT: retq
@ -249,7 +249,7 @@ define <4 x float> @combine_vec_fcopysign_fptrunc_sgn(<4 x float> %x, <4 x doubl
; SSE-NEXT: movaps {{.*#+}} xmm5
; SSE-NEXT: andps %xmm5, %xmm0
; SSE-NEXT: cvtsd2ss %xmm1, %xmm6
; SSE-NEXT: movaps {{.*#+}} xmm4 = [-0,-0,-0,-0]
; SSE-NEXT: movaps {{.*#+}} xmm4 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; SSE-NEXT: andps %xmm4, %xmm6
; SSE-NEXT: orps %xmm6, %xmm0
; SSE-NEXT: movshdup {{.*#+}} xmm6 = xmm3[1,1,3,3]
@ -282,7 +282,7 @@ define <4 x float> @combine_vec_fcopysign_fptrunc_sgn(<4 x float> %x, <4 x doubl
; AVX-NEXT: vbroadcastss {{.*}}(%rip), %xmm2
; AVX-NEXT: vandpd %xmm2, %xmm0, %xmm0
; AVX-NEXT: vcvtpd2ps %ymm1, %xmm1
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
; AVX-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; AVX-NEXT: vandpd %xmm2, %xmm1, %xmm1
; AVX-NEXT: vorpd %xmm1, %xmm0, %xmm0
; AVX-NEXT: vzeroupper

8
test/CodeGen/X86/cvtv2f32.ll
View File

@ -44,7 +44,7 @@ define <2 x float> @uitofp_2i32_buildvector_cvt(i32 %x, i32 %y, <2 x float> %v)
; X32-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; X32-NEXT: movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
; X32-NEXT: unpcklpd {{.*#+}} xmm2 = xmm2[0],xmm1[0]
; X32-NEXT: movapd {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
; X32-NEXT: movapd {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; X32-NEXT: orpd %xmm1, %xmm2
; X32-NEXT: subpd %xmm1, %xmm2
; X32-NEXT: cvtpd2ps %xmm2, %xmm1
@ -56,7 +56,7 @@ define <2 x float> @uitofp_2i32_buildvector_cvt(i32 %x, i32 %y, <2 x float> %v)
; X64-NEXT: movd %esi, %xmm1
; X64-NEXT: movd %edi, %xmm2
; X64-NEXT: punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm1[0]
; X64-NEXT: movdqa {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
; X64-NEXT: movdqa {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; X64-NEXT: por %xmm1, %xmm2
; X64-NEXT: subpd %xmm1, %xmm2
; X64-NEXT: cvtpd2ps %xmm2, %xmm1
@ -74,7 +74,7 @@ define <2 x float> @uitofp_2i32_legalized(<2 x i32> %in, <2 x float> %v) {
; X32: # %bb.0:
; X32-NEXT: xorps %xmm2, %xmm2
; X32-NEXT: blendps {{.*#+}} xmm2 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
; X32-NEXT: movaps {{.*#+}} xmm0 = [4503599627370496,4503599627370496]
; X32-NEXT: movaps {{.*#+}} xmm0 = [4.503599627370496E+15,4.503599627370496E+15]
; X32-NEXT: orps %xmm0, %xmm2
; X32-NEXT: subpd %xmm0, %xmm2
; X32-NEXT: cvtpd2ps %xmm2, %xmm0
@ -85,7 +85,7 @@ define <2 x float> @uitofp_2i32_legalized(<2 x i32> %in, <2 x float> %v) {
; X64: # %bb.0:
; X64-NEXT: xorps %xmm2, %xmm2
; X64-NEXT: blendps {{.*#+}} xmm2 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
; X64-NEXT: movaps {{.*#+}} xmm0 = [4503599627370496,4503599627370496]
; X64-NEXT: movaps {{.*#+}} xmm0 = [4.503599627370496E+15,4.503599627370496E+15]
; X64-NEXT: orps %xmm0, %xmm2
; X64-NEXT: subpd %xmm0, %xmm2
; X64-NEXT: cvtpd2ps %xmm2, %xmm0

2
test/CodeGen/X86/fma-fneg-combine.ll
View File

@ -141,7 +141,7 @@ define <4 x float> @test11(<4 x float> %a, <4 x float> %b, <4 x float> %c, i8 ze
;
; KNL-LABEL: test11:
; KNL: # %bb.0: # %entry
; KNL-NEXT: vbroadcastss {{.*#+}} xmm3 = [-0,-0,-0,-0]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; KNL-NEXT: vxorps %xmm3, %xmm2, %xmm3
; KNL-NEXT: vfmsub213ss {{.*#+}} xmm0 = (xmm1 * xmm0) - xmm2
; KNL-NEXT: kmovw %edi, %k1

8
test/CodeGen/X86/fma-intrinsics-fast-isel.ll
View File

@ -160,7 +160,7 @@ entry:
define <4 x float> @test_mm_fnmsub_ps(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
; CHECK-LABEL: test_mm_fnmsub_ps:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vmovaps {{.*#+}} xmm3 = [-0,-0,-0,-0]
; CHECK-NEXT: vmovaps {{.*#+}} xmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vxorps %xmm3, %xmm0, %xmm4
; CHECK-NEXT: vxorps %xmm3, %xmm2, %xmm0
; CHECK-NEXT: vfmadd231ps {{.*#+}} xmm0 = (xmm1 * xmm4) + xmm0
@ -175,7 +175,7 @@ entry:
define <2 x double> @test_mm_fnmsub_pd(<2 x double> %a, <2 x double> %b, <2 x double> %c) {
; CHECK-LABEL: test_mm_fnmsub_pd:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vmovapd {{.*#+}} xmm3 = [-0,-0]
; CHECK-NEXT: vmovapd {{.*#+}} xmm3 = [-0.0E+0,-0.0E+0]
; CHECK-NEXT: vxorpd %xmm3, %xmm0, %xmm4
; CHECK-NEXT: vxorpd %xmm3, %xmm2, %xmm0
; CHECK-NEXT: vfmadd231pd {{.*#+}} xmm0 = (xmm1 * xmm4) + xmm0
@ -342,7 +342,7 @@ entry:
define <8 x float> @test_mm256_fnmsub_ps(<8 x float> %a, <8 x float> %b, <8 x float> %c) {
; CHECK-LABEL: test_mm256_fnmsub_ps:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vmovaps {{.*#+}} ymm3 = [-0,-0,-0,-0,-0,-0,-0,-0]
; CHECK-NEXT: vmovaps {{.*#+}} ymm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vxorps %ymm3, %ymm0, %ymm4
; CHECK-NEXT: vxorps %ymm3, %ymm2, %ymm0
; CHECK-NEXT: vfmadd231ps {{.*#+}} ymm0 = (ymm1 * ymm4) + ymm0
@ -357,7 +357,7 @@ entry:
define <4 x double> @test_mm256_fnmsub_pd(<4 x double> %a, <4 x double> %b, <4 x double> %c) {
; CHECK-LABEL: test_mm256_fnmsub_pd:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vmovapd {{.*#+}} ymm3 = [-0,-0,-0,-0]
; CHECK-NEXT: vmovapd {{.*#+}} ymm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: vxorpd %ymm3, %ymm0, %ymm4
; CHECK-NEXT: vxorpd %ymm3, %ymm2, %ymm0
; CHECK-NEXT: vfmadd231pd {{.*#+}} ymm0 = (ymm1 * ymm4) + ymm0

60
test/CodeGen/X86/fma_patterns.ll
View File

@ -791,21 +791,21 @@ define <4 x float> @test_v4f32_mul_y_add_x_negone_undefs(<4 x float> %x, <4 x fl
define <4 x float> @test_v4f32_mul_sub_one_x_y(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX512-INFS-NEXT: retq
@ -832,21 +832,21 @@ define <4 x float> @test_v4f32_mul_sub_one_x_y(<4 x float> %x, <4 x float> %y) {
define <4 x float> @test_v4f32_mul_y_sub_one_x(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX512-INFS-NEXT: retq
@ -873,21 +873,21 @@ define <4 x float> @test_v4f32_mul_y_sub_one_x(<4 x float> %x, <4 x float> %y) {
define <4 x float> @test_v4f32_mul_y_sub_one_x_undefs(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_y_sub_one_x_undefs:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <1,u,1,1>
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <1.0E+0,u,1.0E+0,1.0E+0>
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_one_x_undefs:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <1,u,1,1>
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <1.0E+0,u,1.0E+0,1.0E+0>
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_one_x_undefs:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX512-INFS-NEXT: retq
@ -914,21 +914,21 @@ define <4 x float> @test_v4f32_mul_y_sub_one_x_undefs(<4 x float> %x, <4 x float
define <4 x float> @test_v4f32_mul_sub_negone_x_y(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1,-1,-1,-1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX512-INFS-NEXT: retq
@ -955,21 +955,21 @@ define <4 x float> @test_v4f32_mul_sub_negone_x_y(<4 x float> %x, <4 x float> %y
define <4 x float> @test_v4f32_mul_y_sub_negone_x(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1,-1,-1,-1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX512-INFS-NEXT: retq
@ -996,21 +996,21 @@ define <4 x float> @test_v4f32_mul_y_sub_negone_x(<4 x float> %x, <4 x float> %y
define <4 x float> @test_v4f32_mul_y_sub_negone_x_undefs(<4 x float> %x, <4 x float> %y) {
; FMA-INFS-LABEL: test_v4f32_mul_y_sub_negone_x_undefs:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <-1,-1,u,-1>
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <-1.0E+0,-1.0E+0,u,-1.0E+0>
; FMA-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA-INFS-NEXT: retq
;
; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_negone_x_undefs:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <-1,-1,u,-1>
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm2 = <-1.0E+0,-1.0E+0,u,-1.0E+0>
; FMA4-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; FMA4-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; FMA4-INFS-NEXT: retq
;
; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_negone_x_undefs:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1,-1,-1,-1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm2 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX512-INFS-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX512-INFS-NEXT: retq
@ -1318,7 +1318,7 @@ define float @test_f32_interp(float %x, float %y, float %t) {
define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float> %t) {
; FMA-INFS-LABEL: test_v4f32_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubps %xmm2, %xmm3, %xmm3
; FMA-INFS-NEXT: vmulps %xmm3, %xmm1, %xmm1
; FMA-INFS-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
@ -1326,7 +1326,7 @@ define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float
;
; FMA4-INFS-LABEL: test_v4f32_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %xmm2, %xmm3, %xmm3
; FMA4-INFS-NEXT: vmulps %xmm3, %xmm1, %xmm1
; FMA4-INFS-NEXT: vfmaddps %xmm1, %xmm2, %xmm0, %xmm0
@ -1334,7 +1334,7 @@ define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float
;
; AVX512-INFS-LABEL: test_v4f32_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm3 = [1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubps %xmm2, %xmm3, %xmm3
; AVX512-INFS-NEXT: vmulps %xmm3, %xmm1, %xmm1
; AVX512-INFS-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
@ -1367,7 +1367,7 @@ define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float
define <8 x float> @test_v8f32_interp(<8 x float> %x, <8 x float> %y, <8 x float> %t) {
; FMA-INFS-LABEL: test_v8f32_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; FMA-INFS-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-INFS-NEXT: vsubps %ymm2, %ymm3, %ymm3
; FMA-INFS-NEXT: vmulps %ymm3, %ymm1, %ymm1
; FMA-INFS-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
@ -1375,7 +1375,7 @@ define <8 x float> @test_v8f32_interp(<8 x float> %x, <8 x float> %y, <8 x float
;
; FMA4-INFS-LABEL: test_v8f32_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; FMA4-INFS-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]
; FMA4-INFS-NEXT: vsubps %ymm2, %ymm3, %ymm3
; FMA4-INFS-NEXT: vmulps %ymm3, %ymm1, %ymm1
; FMA4-INFS-NEXT: vfmaddps %ymm1, %ymm2, %ymm0, %ymm0
@ -1383,7 +1383,7 @@ define <8 x float> @test_v8f32_interp(<8 x float> %x, <8 x float> %y, <8 x float
;
; AVX512-INFS-LABEL: test_v8f32_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX512-INFS-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]
; AVX512-INFS-NEXT: vsubps %ymm2, %ymm3, %ymm3
; AVX512-INFS-NEXT: vmulps %ymm3, %ymm1, %ymm1
; AVX512-INFS-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
@ -1465,7 +1465,7 @@ define double @test_f64_interp(double %x, double %y, double %t) {
define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x double> %t) {
; FMA-INFS-LABEL: test_v2f64_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubpd %xmm2, %xmm3, %xmm3
; FMA-INFS-NEXT: vmulpd %xmm3, %xmm1, %xmm1
; FMA-INFS-NEXT: vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
@ -1473,7 +1473,7 @@ define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x do
;
; FMA4-INFS-LABEL: test_v2f64_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubpd %xmm2, %xmm3, %xmm3
; FMA4-INFS-NEXT: vmulpd %xmm3, %xmm1, %xmm1
; FMA4-INFS-NEXT: vfmaddpd %xmm1, %xmm2, %xmm0, %xmm0
@ -1481,7 +1481,7 @@ define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x do
;
; AVX512-INFS-LABEL: test_v2f64_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1,1]
; AVX512-INFS-NEXT: vmovapd {{.*#+}} xmm3 = [1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubpd %xmm2, %xmm3, %xmm3
; AVX512-INFS-NEXT: vmulpd %xmm3, %xmm1, %xmm1
; AVX512-INFS-NEXT: vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
@ -1514,7 +1514,7 @@ define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x do
define <4 x double> @test_v4f64_interp(<4 x double> %x, <4 x double> %y, <4 x double> %t) {
; FMA-INFS-LABEL: test_v4f64_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm3 = [1,1,1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm2, %ymm3, %ymm3
; FMA-INFS-NEXT: vmulpd %ymm3, %ymm1, %ymm1
; FMA-INFS-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
@ -1522,7 +1522,7 @@ define <4 x double> @test_v4f64_interp(<4 x double> %x, <4 x double> %y, <4 x do
;
; FMA4-INFS-LABEL: test_v4f64_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm3 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm2, %ymm3, %ymm3
; FMA4-INFS-NEXT: vmulpd %ymm3, %ymm1, %ymm1
; FMA4-INFS-NEXT: vfmaddpd %ymm1, %ymm2, %ymm0, %ymm0
@ -1530,7 +1530,7 @@ define <4 x double> @test_v4f64_interp(<4 x double> %x, <4 x double> %y, <4 x do
;
; AVX512-INFS-LABEL: test_v4f64_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} ymm3 = [1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} ymm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-INFS-NEXT: vsubpd %ymm2, %ymm3, %ymm3
; AVX512-INFS-NEXT: vmulpd %ymm3, %ymm1, %ymm1
; AVX512-INFS-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1

72
test/CodeGen/X86/fma_patterns_wide.ll
View File

@ -259,7 +259,7 @@ define <8 x double> @test_8f64_fmsub_load(<8 x double>* %a0, <8 x double> %a1, <
define <16 x float> @test_v16f32_mul_add_x_one_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_add_x_one_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [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-INFS-NEXT: vaddps %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vaddps %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -268,7 +268,7 @@ define <16 x float> @test_v16f32_mul_add_x_one_y(<16 x float> %x, <16 x float> %
;
; FMA4-INFS-LABEL: test_v16f32_mul_add_x_one_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vaddps %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vaddps %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -305,7 +305,7 @@ define <16 x float> @test_v16f32_mul_add_x_one_y(<16 x float> %x, <16 x float> %
define <8 x double> @test_v8f64_mul_y_add_x_one(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_add_x_one:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vaddpd %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vaddpd %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -314,7 +314,7 @@ define <8 x double> @test_v8f64_mul_y_add_x_one(<8 x double> %x, <8 x double> %y
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_add_x_one:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vaddpd %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vaddpd %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -351,7 +351,7 @@ define <8 x double> @test_v8f64_mul_y_add_x_one(<8 x double> %x, <8 x double> %y
define <16 x float> @test_v16f32_mul_add_x_negone_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_add_x_negone_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-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-INFS-NEXT: vaddps %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vaddps %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -360,7 +360,7 @@ define <16 x float> @test_v16f32_mul_add_x_negone_y(<16 x float> %x, <16 x float
;
; FMA4-INFS-LABEL: test_v16f32_mul_add_x_negone_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vaddps %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vaddps %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -397,7 +397,7 @@ define <16 x float> @test_v16f32_mul_add_x_negone_y(<16 x float> %x, <16 x float
define <8 x double> @test_v8f64_mul_y_add_x_negone(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_add_x_negone:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA-INFS-NEXT: vaddpd %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vaddpd %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -406,7 +406,7 @@ define <8 x double> @test_v8f64_mul_y_add_x_negone(<8 x double> %x, <8 x double>
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_add_x_negone:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vaddpd %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vaddpd %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -443,7 +443,7 @@ define <8 x double> @test_v8f64_mul_y_add_x_negone(<8 x double> %x, <8 x double>
define <16 x float> @test_v16f32_mul_sub_one_x_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [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-INFS-NEXT: vsubps %ymm1, %ymm4, %ymm1
; FMA-INFS-NEXT: vsubps %ymm0, %ymm4, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -452,7 +452,7 @@ define <16 x float> @test_v16f32_mul_sub_one_x_y(<16 x float> %x, <16 x float> %
;
; FMA4-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %ymm1, %ymm4, %ymm1
; FMA4-INFS-NEXT: vsubps %ymm0, %ymm4, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -461,7 +461,7 @@ define <16 x float> @test_v16f32_mul_sub_one_x_y(<16 x float> %x, <16 x float> %
;
; AVX512-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; AVX512-INFS-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-INFS-NEXT: vsubps %zmm0, %zmm2, %zmm0
; AVX512-INFS-NEXT: vmulps %zmm1, %zmm0, %zmm0
; AVX512-INFS-NEXT: retq
@ -490,7 +490,7 @@ define <16 x float> @test_v16f32_mul_sub_one_x_y(<16 x float> %x, <16 x float> %
define <8 x double> @test_v8f64_mul_y_sub_one_x(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm1, %ymm4, %ymm1
; FMA-INFS-NEXT: vsubpd %ymm0, %ymm4, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -499,7 +499,7 @@ define <8 x double> @test_v8f64_mul_y_sub_one_x(<8 x double> %x, <8 x double> %y
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm1, %ymm4, %ymm1
; FMA4-INFS-NEXT: vsubpd %ymm0, %ymm4, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -508,7 +508,7 @@ define <8 x double> @test_v8f64_mul_y_sub_one_x(<8 x double> %x, <8 x double> %y
;
; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} 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]
; AVX512-INFS-NEXT: vsubpd %zmm0, %zmm2, %zmm0
; AVX512-INFS-NEXT: vmulpd %zmm0, %zmm1, %zmm0
; AVX512-INFS-NEXT: retq
@ -537,7 +537,7 @@ define <8 x double> @test_v8f64_mul_y_sub_one_x(<8 x double> %x, <8 x double> %y
define <16 x float> @test_v16f32_mul_sub_negone_x_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-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-INFS-NEXT: vsubps %ymm1, %ymm4, %ymm1
; FMA-INFS-NEXT: vsubps %ymm0, %ymm4, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -546,7 +546,7 @@ define <16 x float> @test_v16f32_mul_sub_negone_x_y(<16 x float> %x, <16 x float
;
; FMA4-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubps %ymm1, %ymm4, %ymm1
; FMA4-INFS-NEXT: vsubps %ymm0, %ymm4, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -555,7 +555,7 @@ define <16 x float> @test_v16f32_mul_sub_negone_x_y(<16 x float> %x, <16 x float
;
; AVX512-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} zmm2 = [-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1]
; AVX512-INFS-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-INFS-NEXT: vsubps %zmm0, %zmm2, %zmm0
; AVX512-INFS-NEXT: vmulps %zmm1, %zmm0, %zmm0
; AVX512-INFS-NEXT: retq
@ -584,7 +584,7 @@ define <16 x float> @test_v16f32_mul_sub_negone_x_y(<16 x float> %x, <16 x float
define <8 x double> @test_v8f64_mul_y_sub_negone_x(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm1, %ymm4, %ymm1
; FMA-INFS-NEXT: vsubpd %ymm0, %ymm4, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -593,7 +593,7 @@ define <8 x double> @test_v8f64_mul_y_sub_negone_x(<8 x double> %x, <8 x double>
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm1, %ymm4, %ymm1
; FMA4-INFS-NEXT: vsubpd %ymm0, %ymm4, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -602,7 +602,7 @@ define <8 x double> @test_v8f64_mul_y_sub_negone_x(<8 x double> %x, <8 x double>
;
; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} zmm2 = [-1,-1,-1,-1,-1,-1,-1,-1]
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} 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]
; AVX512-INFS-NEXT: vsubpd %zmm0, %zmm2, %zmm0
; AVX512-INFS-NEXT: vmulpd %zmm0, %zmm1, %zmm0
; AVX512-INFS-NEXT: retq
@ -631,7 +631,7 @@ define <8 x double> @test_v8f64_mul_y_sub_negone_x(<8 x double> %x, <8 x double>
define <16 x float> @test_v16f32_mul_sub_x_one_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_sub_x_one_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [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-INFS-NEXT: vsubps %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vsubps %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -640,7 +640,7 @@ define <16 x float> @test_v16f32_mul_sub_x_one_y(<16 x float> %x, <16 x float> %
;
; FMA4-INFS-LABEL: test_v16f32_mul_sub_x_one_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vsubps %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -677,7 +677,7 @@ define <16 x float> @test_v16f32_mul_sub_x_one_y(<16 x float> %x, <16 x float> %
define <8 x double> @test_v8f64_mul_y_sub_x_one(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_sub_x_one:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vsubpd %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -686,7 +686,7 @@ define <8 x double> @test_v8f64_mul_y_sub_x_one(<8 x double> %x, <8 x double> %y
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_x_one:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vsubpd %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -723,7 +723,7 @@ define <8 x double> @test_v8f64_mul_y_sub_x_one(<8 x double> %x, <8 x double> %y
define <16 x float> @test_v16f32_mul_sub_x_negone_y(<16 x float> %x, <16 x float> %y) {
; FMA-INFS-LABEL: test_v16f32_mul_sub_x_negone_y:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-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-INFS-NEXT: vsubps %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vsubps %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -732,7 +732,7 @@ define <16 x float> @test_v16f32_mul_sub_x_negone_y(<16 x float> %x, <16 x float
;
; FMA4-INFS-LABEL: test_v16f32_mul_sub_x_negone_y:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubps %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vsubps %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulps %ymm2, %ymm0, %ymm0
@ -769,7 +769,7 @@ define <16 x float> @test_v16f32_mul_sub_x_negone_y(<16 x float> %x, <16 x float
define <8 x double> @test_v8f64_mul_y_sub_x_negone(<8 x double> %x, <8 x double> %y) {
; FMA-INFS-LABEL: test_v8f64_mul_y_sub_x_negone:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm4, %ymm1, %ymm1
; FMA-INFS-NEXT: vsubpd %ymm4, %ymm0, %ymm0
; FMA-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -778,7 +778,7 @@ define <8 x double> @test_v8f64_mul_y_sub_x_negone(<8 x double> %x, <8 x double>
;
; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_x_negone:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm4 = [-1.0E+0,-1.0E+0,-1.0E+0,-1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm4, %ymm1, %ymm1
; FMA4-INFS-NEXT: vsubpd %ymm4, %ymm0, %ymm0
; FMA4-INFS-NEXT: vmulpd %ymm0, %ymm2, %ymm0
@ -819,7 +819,7 @@ define <8 x double> @test_v8f64_mul_y_sub_x_negone(<8 x double> %x, <8 x double>
define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x float> %t) {
; FMA-INFS-LABEL: test_v16f32_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm6 = [1,1,1,1,1,1,1,1]
; FMA-INFS-NEXT: vmovaps {{.*#+}} ymm6 = [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-INFS-NEXT: vsubps %ymm4, %ymm6, %ymm7
; FMA-INFS-NEXT: vsubps %ymm5, %ymm6, %ymm6
; FMA-INFS-NEXT: vmulps %ymm6, %ymm3, %ymm3
@ -830,7 +830,7 @@ define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x
;
; FMA4-INFS-LABEL: test_v16f32_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm6 = [1,1,1,1,1,1,1,1]
; FMA4-INFS-NEXT: vmovaps {{.*#+}} ymm6 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubps %ymm4, %ymm6, %ymm7
; FMA4-INFS-NEXT: vsubps %ymm5, %ymm6, %ymm6
; FMA4-INFS-NEXT: vmulps %ymm6, %ymm3, %ymm3
@ -841,7 +841,7 @@ define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x
;
; AVX512-INFS-LABEL: test_v16f32_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} zmm3 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastss {{.*#+}} zmm3 = [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-INFS-NEXT: vsubps %zmm2, %zmm3, %zmm3
; AVX512-INFS-NEXT: vmulps %zmm3, %zmm1, %zmm1
; AVX512-INFS-NEXT: vfmadd213ps {{.*#+}} zmm0 = (zmm2 * zmm0) + zmm1
@ -878,7 +878,7 @@ define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x
define <8 x double> @test_v8f64_interp(<8 x double> %x, <8 x double> %y, <8 x double> %t) {
; FMA-INFS-LABEL: test_v8f64_interp:
; FMA-INFS: # %bb.0:
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm6 = [1,1,1,1]
; FMA-INFS-NEXT: vmovapd {{.*#+}} ymm6 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-INFS-NEXT: vsubpd %ymm4, %ymm6, %ymm7
; FMA-INFS-NEXT: vsubpd %ymm5, %ymm6, %ymm6
; FMA-INFS-NEXT: vmulpd %ymm6, %ymm3, %ymm3
@ -889,7 +889,7 @@ define <8 x double> @test_v8f64_interp(<8 x double> %x, <8 x double> %y, <8 x do
;
; FMA4-INFS-LABEL: test_v8f64_interp:
; FMA4-INFS: # %bb.0:
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm6 = [1,1,1,1]
; FMA4-INFS-NEXT: vmovapd {{.*#+}} ymm6 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA4-INFS-NEXT: vsubpd %ymm4, %ymm6, %ymm7
; FMA4-INFS-NEXT: vsubpd %ymm5, %ymm6, %ymm6
; FMA4-INFS-NEXT: vmulpd %ymm6, %ymm3, %ymm3
@ -900,7 +900,7 @@ define <8 x double> @test_v8f64_interp(<8 x double> %x, <8 x double> %y, <8 x do
;
; AVX512-INFS-LABEL: test_v8f64_interp:
; AVX512-INFS: # %bb.0:
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} zmm3 = [1,1,1,1,1,1,1,1]
; AVX512-INFS-NEXT: vbroadcastsd {{.*#+}} zmm3 = [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-INFS-NEXT: vsubpd %zmm2, %zmm3, %zmm3
; AVX512-INFS-NEXT: vmulpd %zmm3, %zmm1, %zmm1
; AVX512-INFS-NEXT: vfmadd213pd {{.*#+}} zmm0 = (zmm2 * zmm0) + zmm1
@ -1143,7 +1143,7 @@ define <8 x double> @test_v8f64_fneg_fmul_no_nsz(<8 x double> %x, <8 x double> %
; FMA: # %bb.0:
; FMA-NEXT: vmulpd %ymm3, %ymm1, %ymm1
; FMA-NEXT: vmulpd %ymm2, %ymm0, %ymm0
; FMA-NEXT: vmovapd {{.*#+}} ymm2 = [-0,-0,-0,-0]
; FMA-NEXT: vmovapd {{.*#+}} ymm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; FMA-NEXT: vxorpd %ymm2, %ymm0, %ymm0
; FMA-NEXT: vxorpd %ymm2, %ymm1, %ymm1
; FMA-NEXT: retq
@ -1152,7 +1152,7 @@ define <8 x double> @test_v8f64_fneg_fmul_no_nsz(<8 x double> %x, <8 x double> %
; FMA4: # %bb.0:
; FMA4-NEXT: vmulpd %ymm3, %ymm1, %ymm1
; FMA4-NEXT: vmulpd %ymm2, %ymm0, %ymm0
; FMA4-NEXT: vmovapd {{.*#+}} ymm2 = [-0,-0,-0,-0]
; FMA4-NEXT: vmovapd {{.*#+}} ymm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; FMA4-NEXT: vxorpd %ymm2, %ymm0, %ymm0
; FMA4-NEXT: vxorpd %ymm2, %ymm1, %ymm1
; FMA4-NEXT: retq

4
test/CodeGen/X86/fmul-combines.ll
View File

@ -61,7 +61,7 @@ define <4 x float> @fmul2_v4f32_undef(<4 x float> %x) {
define <4 x float> @constant_fold_fmul_v4f32(<4 x float> %x) {
; CHECK-LABEL: constant_fold_fmul_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [8,8,8,8]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [8.0E+0,8.0E+0,8.0E+0,8.0E+0]
; CHECK-NEXT: retq
%y = fmul <4 x float> <float 4.0, float 4.0, float 4.0, float 4.0>, <float 2.0, float 2.0, float 2.0, float 2.0>
ret <4 x float> %y
@ -70,7 +70,7 @@ define <4 x float> @constant_fold_fmul_v4f32(<4 x float> %x) {
define <4 x float> @constant_fold_fmul_v4f32_undef(<4 x float> %x) {
; CHECK-LABEL: constant_fold_fmul_v4f32_undef:
; CHECK: # %bb.0:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [8,NaN,8,NaN]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [8.0E+0,NaN,8.0E+0,NaN]
; CHECK-NEXT: retq
%y = fmul <4 x float> <float 4.0, float undef, float 4.0, float 4.0>, <float 2.0, float 2.0, float 2.0, float undef>
ret <4 x float> %y

2
test/CodeGen/X86/fold-vector-trunc-sitofp.ll
View File

@ -7,7 +7,7 @@
define <4 x float> @test1() {
; CHECK-LABEL: test1:
; CHECK: # %bb.0:
; CHECK-NEXT: vmovaps {{.*#+}} xmm0 = [-1,0,-1,0]
; CHECK-NEXT: vmovaps {{.*#+}} xmm0 = [-1.0E+0,0.0E+0,-1.0E+0,0.0E+0]
; CHECK-NEXT: ret{{[l|q]}}
%1 = trunc <4 x i3> <i3 -1, i3 -22, i3 7, i3 8> to <4 x i1>
%2 = sitofp <4 x i1> %1 to <4 x float>

68
test/CodeGen/X86/insert-into-constant-vector.ll
View File

@ -167,40 +167,40 @@ define <4 x float> @elt1_v4f32(float %x) {
; X32SSE2-LABEL: elt1_v4f32:
; X32SSE2: # %bb.0:
; X32SSE2-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; X32SSE2-NEXT: movaps {{.*#+}} xmm1 = <42,u,2,3>
; X32SSE2-NEXT: movaps {{.*#+}} xmm1 = <4.2E+1,u,2.0E+0,3.0E+0>
; X32SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[0,0]
; X32SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
; X32SSE2-NEXT: retl
;
; X64SSE2-LABEL: elt1_v4f32:
; X64SSE2: # %bb.0:
; X64SSE2-NEXT: movaps {{.*#+}} xmm1 = <42,u,2,3>
; X64SSE2-NEXT: movaps {{.*#+}} xmm1 = <4.2E+1,u,2.0E+0,3.0E+0>
; X64SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[0,0]
; X64SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
; X64SSE2-NEXT: retq
;
; X32SSE4-LABEL: elt1_v4f32:
; X32SSE4: # %bb.0:
; X32SSE4-NEXT: movaps {{.*#+}} xmm0 = <42,u,2,3>
; X32SSE4-NEXT: movaps {{.*#+}} xmm0 = <4.2E+1,u,2.0E+0,3.0E+0>
; X32SSE4-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
; X32SSE4-NEXT: retl
;
; X64SSE4-LABEL: elt1_v4f32:
; X64SSE4: # %bb.0:
; X64SSE4-NEXT: movaps {{.*#+}} xmm1 = <42,u,2,3>
; X64SSE4-NEXT: movaps {{.*#+}} xmm1 = <4.2E+1,u,2.0E+0,3.0E+0>
; X64SSE4-NEXT: insertps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[2,3]
; X64SSE4-NEXT: movaps %xmm1, %xmm0
; X64SSE4-NEXT: retq
;
; X32AVX-LABEL: elt1_v4f32:
; X32AVX: # %bb.0:
; X32AVX-NEXT: vmovaps {{.*#+}} xmm0 = <42,u,2,3>
; X32AVX-NEXT: vmovaps {{.*#+}} xmm0 = <4.2E+1,u,2.0E+0,3.0E+0>
; X32AVX-NEXT: vinsertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
; X32AVX-NEXT: retl
;
; X64AVX-LABEL: elt1_v4f32:
; X64AVX: # %bb.0:
; X64AVX-NEXT: vmovaps {{.*#+}} xmm1 = <42,u,2,3>
; X64AVX-NEXT: vmovaps {{.*#+}} xmm1 = <4.2E+1,u,2.0E+0,3.0E+0>
; X64AVX-NEXT: vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[2,3]
; X64AVX-NEXT: retq
%ins = insertelement <4 x float> <float 42.0, float 1.0, float 2.0, float 3.0>, float %x, i32 1
@ -210,26 +210,26 @@ define <4 x float> @elt1_v4f32(float %x) {
define <2 x double> @elt1_v2f64(double %x) {
; X32SSE-LABEL: elt1_v2f64:
; X32SSE: # %bb.0:
; X32SSE-NEXT: movapd {{.*#+}} xmm0 = <42,u>
; X32SSE-NEXT: movapd {{.*#+}} xmm0 = <4.2E+1,u>
; X32SSE-NEXT: movhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
; X32SSE-NEXT: retl
;
; X64SSE-LABEL: elt1_v2f64:
; X64SSE: # %bb.0:
; X64SSE-NEXT: movaps {{.*#+}} xmm1 = <42,u>
; X64SSE-NEXT: movaps {{.*#+}} xmm1 = <4.2E+1,u>
; X64SSE-NEXT: movlhps {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; X64SSE-NEXT: movaps %xmm1, %xmm0
; X64SSE-NEXT: retq
;
; X32AVX-LABEL: elt1_v2f64:
; X32AVX: # %bb.0:
; X32AVX-NEXT: vmovapd {{.*#+}} xmm0 = <42,u>
; X32AVX-NEXT: vmovapd {{.*#+}} xmm0 = <4.2E+1,u>
; X32AVX-NEXT: vmovhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
; X32AVX-NEXT: retl
;
; X64AVX-LABEL: elt1_v2f64:
; X64AVX: # %bb.0:
; X64AVX-NEXT: vmovaps {{.*#+}} xmm1 = <42,u>
; X64AVX-NEXT: vmovaps {{.*#+}} xmm1 = <4.2E+1,u>
; X64AVX-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; X64AVX-NEXT: retq
%ins = insertelement <2 x double> <double 42.0, double 1.0>, double %x, i32 1
@ -292,37 +292,37 @@ define <8 x float> @elt6_v8f32(float %x) {
; X32SSE2-LABEL: elt6_v8f32:
; X32SSE2: # %bb.0:
; X32SSE2-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; X32SSE2-NEXT: movaps {{.*#+}} xmm1 = <4,5,u,7>
; X32SSE2-NEXT: movaps {{.*#+}} xmm1 = <4.0E+0,5.0E+0,u,7.0E+0>
; X32SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
; X32SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0,2]
; X32SSE2-NEXT: movaps {{.*#+}} xmm0 = [42,1,2,3]
; X32SSE2-NEXT: movaps {{.*#+}} xmm0 = [4.2E+1,1.0E+0,2.0E+0,3.0E+0]
; X32SSE2-NEXT: retl
;
; X64SSE2-LABEL: elt6_v8f32:
; X64SSE2: # %bb.0:
; X64SSE2-NEXT: movaps {{.*#+}} xmm1 = <4,5,u,7>
; X64SSE2-NEXT: movaps {{.*#+}} xmm1 = <4.0E+0,5.0E+0,u,7.0E+0>
; X64SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
; X64SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0,2]
; X64SSE2-NEXT: movaps {{.*#+}} xmm0 = [42,1,2,3]
; X64SSE2-NEXT: movaps {{.*#+}} xmm0 = [4.2E+1,1.0E+0,2.0E+0,3.0E+0]
; X64SSE2-NEXT: retq
;
; X32SSE4-LABEL: elt6_v8f32:
; X32SSE4: # %bb.0:
; X32SSE4-NEXT: movaps {{.*#+}} xmm1 = <4,5,u,7>
; X32SSE4-NEXT: movaps {{.*#+}} xmm1 = <4.0E+0,5.0E+0,u,7.0E+0>
; X32SSE4-NEXT: insertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
; X32SSE4-NEXT: movaps {{.*#+}} xmm0 = [42,1,2,3]
; X32SSE4-NEXT: movaps {{.*#+}} xmm0 = [4.2E+1,1.0E+0,2.0E+0,3.0E+0]
; X32SSE4-NEXT: retl
;
; X64SSE4-LABEL: elt6_v8f32:
; X64SSE4: # %bb.0:
; X64SSE4-NEXT: movaps {{.*#+}} xmm1 = <4,5,u,7>
; X64SSE4-NEXT: movaps {{.*#+}} xmm1 = <4.0E+0,5.0E+0,u,7.0E+0>
; X64SSE4-NEXT: insertps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0],xmm1[3]
; X64SSE4-NEXT: movaps {{.*#+}} xmm0 = [42,1,2,3]
; X64SSE4-NEXT: movaps {{.*#+}} xmm0 = [4.2E+1,1.0E+0,2.0E+0,3.0E+0]
; X64SSE4-NEXT: retq
;
; X32AVX-LABEL: elt6_v8f32:
; X32AVX: # %bb.0:
; X32AVX-NEXT: vmovaps {{.*#+}} ymm0 = <42,1,2,3,4,5,u,7>
; X32AVX-NEXT: vmovaps {{.*#+}} ymm0 = <4.2E+1,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,u,7.0E+0>
; X32AVX-NEXT: vextractf128 $1, %ymm0, %xmm1
; X32AVX-NEXT: vinsertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
; X32AVX-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm0
@ -330,7 +330,7 @@ define <8 x float> @elt6_v8f32(float %x) {
;
; X64AVX-LABEL: elt6_v8f32:
; X64AVX: # %bb.0:
; X64AVX-NEXT: vmovaps {{.*#+}} ymm1 = <42,1,2,3,4,5,u,7>
; X64AVX-NEXT: vmovaps {{.*#+}} ymm1 = <4.2E+1,1.0E+0,2.0E+0,3.0E+0,4.0E+0,5.0E+0,u,7.0E+0>
; X64AVX-NEXT: vextractf128 $1, %ymm1, %xmm2
; X64AVX-NEXT: vinsertps {{.*#+}} xmm0 = xmm2[0,1],xmm0[0],xmm2[3]
; X64AVX-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
@ -413,49 +413,49 @@ define <8 x i64> @elt5_v8i64(i64 %x) {
define <8 x double> @elt1_v8f64(double %x) {
; X32SSE-LABEL: elt1_v8f64:
; X32SSE: # %bb.0:
; X32SSE-NEXT: movapd {{.*#+}} xmm0 = <42,u>
; X32SSE-NEXT: movapd {{.*#+}} xmm0 = <4.2E+1,u>
; X32SSE-NEXT: movhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
; X32SSE-NEXT: movaps {{.*#+}} xmm1 = [2,3]
; X32SSE-NEXT: movaps {{.*#+}} xmm2 = [4,5]
; X32SSE-NEXT: movaps {{.*#+}} xmm3 = [6,7]
; X32SSE-NEXT: movaps {{.*#+}} xmm1 = [2.0E+0,3.0E+0]
; X32SSE-NEXT: movaps {{.*#+}} xmm2 = [4.0E+0,5.0E+0]
; X32SSE-NEXT: movaps {{.*#+}} xmm3 = [6.0E+0,7.0E+0]
; X32SSE-NEXT: retl
;
; X64SSE-LABEL: elt1_v8f64:
; X64SSE: # %bb.0:
; X64SSE-NEXT: movaps {{.*#+}} xmm4 = <42,u>
; X64SSE-NEXT: movaps {{.*#+}} xmm4 = <4.2E+1,u>
; X64SSE-NEXT: movlhps {{.*#+}} xmm4 = xmm4[0],xmm0[0]
; X64SSE-NEXT: movaps {{.*#+}} xmm1 = [2,3]
; X64SSE-NEXT: movaps {{.*#+}} xmm2 = [4,5]
; X64SSE-NEXT: movaps {{.*#+}} xmm3 = [6,7]
; X64SSE-NEXT: movaps {{.*#+}} xmm1 = [2.0E+0,3.0E+0]
; X64SSE-NEXT: movaps {{.*#+}} xmm2 = [4.0E+0,5.0E+0]
; X64SSE-NEXT: movaps {{.*#+}} xmm3 = [6.0E+0,7.0E+0]
; X64SSE-NEXT: movaps %xmm4, %xmm0
; X64SSE-NEXT: retq
;
; X32AVX2-LABEL: elt1_v8f64:
; X32AVX2: # %bb.0:
; X32AVX2-NEXT: vmovapd {{.*#+}} ymm0 = <42,u,2,3>
; X32AVX2-NEXT: vmovapd {{.*#+}} ymm0 = <4.2E+1,u,2.0E+0,3.0E+0>
; X32AVX2-NEXT: vmovhpd {{.*#+}} xmm1 = xmm0[0],mem[0]
; X32AVX2-NEXT: vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
; X32AVX2-NEXT: vmovaps {{.*#+}} ymm1 = [4,5,6,7]
; X32AVX2-NEXT: vmovaps {{.*#+}} ymm1 = [4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; X32AVX2-NEXT: retl
;
; X64AVX2-LABEL: elt1_v8f64:
; X64AVX2: # %bb.0:
; X64AVX2-NEXT: vmovaps {{.*#+}} ymm1 = <42,u,2,3>
; X64AVX2-NEXT: vmovaps {{.*#+}} ymm1 = <4.2E+1,u,2.0E+0,3.0E+0>
; X64AVX2-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; X64AVX2-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; X64AVX2-NEXT: vmovaps {{.*#+}} ymm1 = [4,5,6,7]
; X64AVX2-NEXT: vmovaps {{.*#+}} ymm1 = [4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; X64AVX2-NEXT: retq
;
; X32AVX512F-LABEL: elt1_v8f64:
; X32AVX512F: # %bb.0:
; X32AVX512F-NEXT: vmovapd {{.*#+}} zmm0 = <42,u,2,3,4,5,6,7>
; X32AVX512F-NEXT: vmovapd {{.*#+}} zmm0 = <4.2E+1,u,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0>
; X32AVX512F-NEXT: vmovhpd {{.*#+}} xmm1 = xmm0[0],mem[0]
; X32AVX512F-NEXT: vinsertf32x4 $0, %xmm1, %zmm0, %zmm0
; X32AVX512F-NEXT: retl
;
; X64AVX512F-LABEL: elt1_v8f64:
; X64AVX512F: # %bb.0:
; X64AVX512F-NEXT: vmovaps {{.*#+}} zmm1 = <42,u,2,3,4,5,6,7>
; X64AVX512F-NEXT: vmovaps {{.*#+}} zmm1 = <4.2E+1,u,2.0E+0,3.0E+0,4.0E+0,5.0E+0,6.0E+0,7.0E+0>
; X64AVX512F-NEXT: vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
; X64AVX512F-NEXT: vinsertf32x4 $0, %xmm0, %zmm1, %zmm0
; X64AVX512F-NEXT: retq

2
test/CodeGen/X86/packss.ll
View File

@ -166,7 +166,7 @@ define <8 x i16> @trunc_ashr_v4i64_demandedelts(<4 x i64> %a0) {
; X86-SSE-NEXT: movsd {{.*#+}} xmm4 = xmm2[0],xmm4[1]
; X86-SSE-NEXT: psrlq $63, %xmm4
; X86-SSE-NEXT: movsd {{.*#+}} xmm0 = xmm4[0],xmm0[1]
; X86-SSE-NEXT: movapd {{.*#+}} xmm2 = [4.9406564584124654E-324,-0]
; X86-SSE-NEXT: movapd {{.*#+}} xmm2 = [4.9406564584124654E-324,-0.0E+0]
; X86-SSE-NEXT: xorpd %xmm2, %xmm0
; X86-SSE-NEXT: psubq %xmm2, %xmm0
; X86-SSE-NEXT: psrlq $63, %xmm3

4
test/CodeGen/X86/pow.ll
View File

@ -56,11 +56,11 @@ define <4 x float> @pow_v4f32_one_fourth_fmf(<4 x float> %x) nounwind {
; CHECK-NEXT: rsqrtps %xmm0, %xmm1
; CHECK-NEXT: movaps %xmm0, %xmm2
; CHECK-NEXT: mulps %xmm1, %xmm2
; CHECK-NEXT: movaps {{.*#+}} xmm3 = [-0.5,-0.5,-0.5,-0.5]
; CHECK-NEXT: movaps {{.*#+}} xmm3 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; CHECK-NEXT: movaps %xmm2, %xmm4
; CHECK-NEXT: mulps %xmm3, %xmm4
; CHECK-NEXT: mulps %xmm1, %xmm2
; CHECK-NEXT: movaps {{.*#+}} xmm1 = [-3,-3,-3,-3]
; CHECK-NEXT: movaps {{.*#+}} xmm1 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; CHECK-NEXT: addps %xmm1, %xmm2
; CHECK-NEXT: mulps %xmm4, %xmm2
; CHECK-NEXT: xorps %xmm4, %xmm4

2
test/CodeGen/X86/pr2656.ll
View File

@ -19,7 +19,7 @@ define void @foo(%struct.anon* byval %p) nounwind {
; CHECK-NEXT: subl $28, %esp
; CHECK-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; CHECK-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; CHECK-NEXT: movaps {{.*#+}} xmm2 = [-0,-0,-0,-0]
; CHECK-NEXT: movaps {{.*#+}} xmm2 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
; CHECK-NEXT: xorps %xmm2, %xmm0
; CHECK-NEXT: cvtss2sd %xmm0, %xmm0
; CHECK-NEXT: xorps %xmm2, %xmm1

4
test/CodeGen/X86/pr38639.ll
View File

@ -4,11 +4,11 @@
define <8 x double> @test(<4 x double> %a, <4 x double> %b) {
; CHECK-LABEL: test:
; CHECK: # %bb.0:
; CHECK-NEXT: vmovaps {{.*#+}} ymm1 = <u,0.82071743224100002,0.82071743224100002,0.82071743224100002>
; CHECK-NEXT: vmovaps {{.*#+}} ymm1 = <u,8.2071743224100002E-1,8.2071743224100002E-1,8.2071743224100002E-1>
; CHECK-NEXT: vblendps {{.*#+}} ymm2 = ymm0[0,1,2,3],ymm1[4,5,6,7]
; CHECK-NEXT: vblendps {{.*#+}} ymm1 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
; CHECK-NEXT: vunpckhpd {{.*#+}} ymm1 = ymm1[1],ymm2[1],ymm1[3],ymm2[3]
; CHECK-NEXT: vmovaps {{.*#+}} xmm2 = [0.82071743224100002,0.82071743224100002]
; CHECK-NEXT: vmovaps {{.*#+}} xmm2 = [8.2071743224100002E-1,8.2071743224100002E-1]
; CHECK-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7]
; CHECK-NEXT: retq
%1 = shufflevector <4 x double> %a, <4 x double> <double undef, double 0x3FEA435134576E1C, double 0x3FEA435134576E1C, double 0x3FEA435134576E1C>, <8 x i32> <i32 6, i32 5, i32 2, i32 3, i32 5, i32 1, i32 3, i32 7>

164
test/CodeGen/X86/recip-fastmath.ll
View File

@ -307,62 +307,62 @@ define float @f32_two_step(float %x) #2 {
define <4 x float> @v4f32_no_estimate(<4 x float> %x) #0 {
; SSE-LABEL: v4f32_no_estimate:
; SSE: # %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: divps %xmm0, %xmm1
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-RECIP-LABEL: v4f32_no_estimate:
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vdivps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v4f32_no_estimate:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1]
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; FMA-RECIP-NEXT: vdivps %xmm0, %xmm1, %xmm0
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v4f32_no_estimate:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [5:0.50]
; BDVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:0.50]
; BDVER2-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [9:9.50]
; BDVER2-NEXT: retq # sched: [5:1.00]
;
; BTVER2-LABEL: v4f32_no_estimate:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [19:19.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v4f32_no_estimate:
; SANDY: # %bb.0:
; SANDY-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [14:14.00]
; SANDY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: v4f32_no_estimate:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [13:7.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; HASWELL-NO-FMA-LABEL: v4f32_no_estimate:
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NO-FMA-NEXT: vdivps %xmm0, %xmm1, %xmm0
; HASWELL-NO-FMA-NEXT: retq
;
; KNL-LABEL: v4f32_no_estimate:
; KNL: # %bb.0:
; KNL-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [13:7.00]
; KNL-NEXT: retq # sched: [7:1.00]
;
; SKX-LABEL: v4f32_no_estimate:
; SKX: # %bb.0:
; SKX-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
; SKX-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SKX-NEXT: vdivps %xmm0, %xmm1, %xmm0 # sched: [11:3.00]
; SKX-NEXT: retq # sched: [7:1.00]
%div = fdiv fast <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
@ -374,7 +374,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: subps %xmm0, %xmm1
; SSE-NEXT: mulps %xmm2, %xmm1
; SSE-NEXT: addps %xmm2, %xmm1
@ -385,7 +385,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX-RECIP-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: vaddps %xmm0, %xmm1, %xmm0
@ -407,7 +407,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
;
; BTVER2-LABEL: v4f32_one_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
@ -419,7 +419,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@ -428,7 +428,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; HASWELL-LABEL: v4f32_one_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; HASWELL-NEXT: retq # sched: [7:1.00]
@ -437,7 +437,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NO-FMA-NEXT: vmulps %xmm1, %xmm0, %xmm0
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NO-FMA-NEXT: vsubps %xmm0, %xmm2, %xmm0
; HASWELL-NO-FMA-NEXT: vmulps %xmm0, %xmm1, %xmm0
; HASWELL-NO-FMA-NEXT: vaddps %xmm0, %xmm1, %xmm0
@ -446,7 +446,7 @@ define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
; KNL-LABEL: v4f32_one_step:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; KNL-NEXT: retq # sched: [7:1.00]
@ -467,7 +467,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: movaps %xmm0, %xmm3
; SSE-NEXT: mulps %xmm2, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm1, %xmm4
; SSE-NEXT: subps %xmm3, %xmm4
; SSE-NEXT: mulps %xmm2, %xmm4
@ -483,7 +483,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm2
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %xmm2, %xmm3, %xmm2
; AVX-RECIP-NEXT: vmulps %xmm2, %xmm1, %xmm2
; AVX-RECIP-NEXT: vaddps %xmm2, %xmm1, %xmm1
@ -496,7 +496,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; FMA-RECIP-LABEL: v4f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %xmm0, %xmm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; 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
@ -507,7 +507,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; BDVER2-LABEL: v4f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:0.50]
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm3 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %xmm1, %xmm3, %xmm1, %xmm1 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm0 # sched: [5:0.50]
@ -516,7 +516,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
;
; BTVER2-LABEL: v4f32_two_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
@ -532,7 +532,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm2, %xmm1, %xmm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
@ -545,7 +545,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; HASWELL-LABEL: v4f32_two_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [5:0.50]
@ -557,7 +557,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %xmm0, %xmm1
; HASWELL-NO-FMA-NEXT: vmulps %xmm1, %xmm0, %xmm2
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm3 = [1,1,1,1]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; HASWELL-NO-FMA-NEXT: vsubps %xmm2, %xmm3, %xmm2
; HASWELL-NO-FMA-NEXT: vmulps %xmm2, %xmm1, %xmm2
; HASWELL-NO-FMA-NEXT: vaddps %xmm2, %xmm1, %xmm1
@ -570,7 +570,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; KNL-LABEL: v4f32_two_step:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [5:0.50]
@ -581,7 +581,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
; SKX-LABEL: v4f32_two_step:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %xmm0, %xmm1 # sched: [4:1.00]
; SKX-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SKX-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SKX-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [4:0.50]
@ -595,7 +595,7 @@ define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
; SSE-LABEL: v8f32_no_estimate:
; SSE: # %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: divps %xmm0, %xmm3
; SSE-NEXT: divps %xmm1, %xmm2
@ -605,55 +605,55 @@ define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
;
; AVX-RECIP-LABEL: v8f32_no_estimate:
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vdivps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v8f32_no_estimate:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm1 = [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: vdivps %ymm0, %ymm1, %ymm0
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v8f32_no_estimate:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:0.50]
; BDVER2-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [9:19.00]
; BDVER2-NEXT: retq # sched: [5:1.00]
;
; BTVER2-LABEL: v8f32_no_estimate:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [38:38.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v8f32_no_estimate:
; SANDY: # %bb.0:
; SANDY-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; SANDY-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [29:28.00]
; SANDY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: v8f32_no_estimate:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; HASWELL-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [21:14.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; HASWELL-NO-FMA-LABEL: v8f32_no_estimate:
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm1 = [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-NO-FMA-NEXT: vdivps %ymm0, %ymm1, %ymm0
; HASWELL-NO-FMA-NEXT: retq
;
; KNL-LABEL: v8f32_no_estimate:
; KNL: # %bb.0:
; KNL-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; KNL-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [21:14.00]
; KNL-NEXT: retq # sched: [7:1.00]
;
; SKX-LABEL: v8f32_no_estimate:
; SKX: # %bb.0:
; SKX-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SKX-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; SKX-NEXT: vdivps %ymm0, %ymm1, %ymm0 # sched: [11:5.00]
; SKX-NEXT: retq # sched: [7:1.00]
%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
@ -665,7 +665,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm4
; SSE-NEXT: mulps %xmm4, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: subps %xmm0, %xmm3
; SSE-NEXT: mulps %xmm4, %xmm3
@ -683,7 +683,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm1, %ymm0
@ -705,7 +705,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
;
; BTVER2-LABEL: v8f32_one_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
@ -717,7 +717,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@ -726,7 +726,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; HASWELL-LABEL: v8f32_one_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; HASWELL-NEXT: retq # sched: [7:1.00]
@ -735,7 +735,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm0, %ymm0
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-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-NO-FMA-NEXT: vsubps %ymm0, %ymm2, %ymm0
; HASWELL-NO-FMA-NEXT: vmulps %ymm0, %ymm1, %ymm0
; HASWELL-NO-FMA-NEXT: vaddps %ymm0, %ymm1, %ymm0
@ -744,7 +744,7 @@ define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
; KNL-LABEL: v8f32_one_step:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; KNL-NEXT: retq # sched: [7:1.00]
@ -766,7 +766,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm3
; SSE-NEXT: movaps %xmm0, %xmm4
; SSE-NEXT: mulps %xmm3, %xmm4
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm1, %xmm5
; SSE-NEXT: subps %xmm4, %xmm5
; SSE-NEXT: mulps %xmm3, %xmm5
@ -794,7 +794,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm2
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm2, %ymm3, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm2
; AVX-RECIP-NEXT: vaddps %ymm2, %ymm1, %ymm1
@ -807,7 +807,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; FMA-RECIP-LABEL: v8f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm1
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; 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
@ -818,7 +818,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; BDVER2-LABEL: v8f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm3 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %ymm1, %ymm3, %ymm1, %ymm1 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm0 # sched: [5:0.50]
@ -827,7 +827,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
;
; BTVER2-LABEL: v8f32_two_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:2.00]
@ -843,7 +843,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm2, %ymm1, %ymm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
@ -856,7 +856,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; HASWELL-LABEL: v8f32_two_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [5:0.50]
@ -868,7 +868,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm1
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm0, %ymm2
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-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-NO-FMA-NEXT: vsubps %ymm2, %ymm3, %ymm2
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm1, %ymm2
; HASWELL-NO-FMA-NEXT: vaddps %ymm2, %ymm1, %ymm1
@ -881,7 +881,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; KNL-LABEL: v8f32_two_step:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; KNL-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [5:0.50]
@ -892,7 +892,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
; SKX-LABEL: v8f32_two_step:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %ymm0, %ymm1 # sched: [4:1.00]
; SKX-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SKX-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] sched: [7:0.50]
; SKX-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [4:0.50]
@ -906,7 +906,7 @@ define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
; SSE-LABEL: v16f32_no_estimate:
; SSE: # %bb.0:
; SSE-NEXT: movaps {{.*#+}} xmm4 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm4, %xmm5
; SSE-NEXT: divps %xmm0, %xmm5
; SSE-NEXT: movaps %xmm4, %xmm6
@ -922,62 +922,62 @@ define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
;
; AVX-RECIP-LABEL: v16f32_no_estimate:
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vdivps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vdivps %ymm1, %ymm2, %ymm1
; AVX-RECIP-NEXT: retq
;
; FMA-RECIP-LABEL: v16f32_no_estimate:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; 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: vdivps %ymm0, %ymm2, %ymm0
; FMA-RECIP-NEXT: vdivps %ymm1, %ymm2, %ymm1
; FMA-RECIP-NEXT: retq
;
; BDVER2-LABEL: v16f32_no_estimate:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [9:19.00]
; BDVER2-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [9:19.00]
; BDVER2-NEXT: retq # sched: [5:1.00]
;
; BTVER2-LABEL: v16f32_no_estimate:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [38:38.00]
; BTVER2-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [38:38.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; SANDY-LABEL: v16f32_no_estimate:
; SANDY: # %bb.0:
; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [29:28.00]
; SANDY-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [29:28.00]
; SANDY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: v16f32_no_estimate:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vdivps %ymm0, %ymm2, %ymm0 # sched: [21:14.00]
; HASWELL-NEXT: vdivps %ymm1, %ymm2, %ymm1 # sched: [21:14.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; HASWELL-NO-FMA-LABEL: v16f32_no_estimate:
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-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-NO-FMA-NEXT: vdivps %ymm0, %ymm2, %ymm0
; HASWELL-NO-FMA-NEXT: vdivps %ymm1, %ymm2, %ymm1
; HASWELL-NO-FMA-NEXT: retq
;
; KNL-LABEL: v16f32_no_estimate:
; KNL: # %bb.0:
; KNL-NEXT: vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [10:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} zmm1 = [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] sched: [10:1.00]
; KNL-NEXT: vdivps %zmm0, %zmm1, %zmm0 # sched: [21:14.00]
; KNL-NEXT: retq # sched: [7:1.00]
;
; SKX-LABEL: v16f32_no_estimate:
; SKX: # %bb.0:
; SKX-NEXT: vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [8:0.50]
; SKX-NEXT: vbroadcastss {{.*#+}} zmm1 = [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] sched: [8:0.50]
; SKX-NEXT: vdivps %zmm0, %zmm1, %zmm0 # sched: [18:10.00]
; SKX-NEXT: retq # sched: [7:1.00]
%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
@ -991,7 +991,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; SSE-NEXT: movaps %xmm0, %xmm5
; SSE-NEXT: rcpps %xmm0, %xmm6
; SSE-NEXT: mulps %xmm6, %xmm5
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm3, %xmm0
; SSE-NEXT: subps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm6, %xmm0
@ -1021,7 +1021,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm0, %ymm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm3, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm2, %ymm0
@ -1035,7 +1035,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; FMA-RECIP-LABEL: v16f32_one_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; 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: vrcpps %ymm1, %ymm2
@ -1046,7 +1046,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; BDVER2-LABEL: v16f32_one_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vrcpps %ymm1, %ymm4 # sched: [5:2.00]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm4, %ymm1, %ymm1 # sched: [5:0.50]
@ -1056,7 +1056,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
;
; BTVER2-LABEL: v16f32_one_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vrcpps %ymm1, %ymm4 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
@ -1073,7 +1073,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
@ -1087,7 +1087,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; HASWELL-LABEL: v16f32_one_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm2 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vrcpps %ymm1, %ymm4 # sched: [11:2.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2 sched: [5:0.50]
@ -1099,7 +1099,7 @@ define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm0, %ymm0
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-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-NO-FMA-NEXT: vsubps %ymm0, %ymm3, %ymm0
; HASWELL-NO-FMA-NEXT: vmulps %ymm0, %ymm2, %ymm0
; HASWELL-NO-FMA-NEXT: vaddps %ymm0, %ymm2, %ymm0
@ -1136,7 +1136,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm0
; SSE-NEXT: movaps %xmm1, %xmm6
; SSE-NEXT: mulps %xmm0, %xmm6
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm3, %xmm7
; SSE-NEXT: subps %xmm6, %xmm7
; SSE-NEXT: mulps %xmm0, %xmm7
@ -1188,7 +1188,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm0, %ymm3
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %ymm3, %ymm4, %ymm3
; AVX-RECIP-NEXT: vmulps %ymm3, %ymm2, %ymm3
; AVX-RECIP-NEXT: vaddps %ymm3, %ymm2, %ymm2
@ -1210,7 +1210,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; FMA-RECIP-LABEL: v16f32_two_step:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; 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
@ -1227,7 +1227,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; BDVER2-LABEL: v16f32_two_step:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm4 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %ymm2, %ymm4, %ymm2, %ymm2 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0 # sched: [5:0.50]
@ -1241,7 +1241,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
;
; BTVER2-LABEL: v16f32_two_step:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm0, %ymm3 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
@ -1266,7 +1266,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm0, %ymm3 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm3, %ymm2, %ymm3 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
@ -1288,7 +1288,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; HASWELL-LABEL: v16f32_two_step:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm2 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vmovaps %ymm2, %ymm4 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2 sched: [5:0.50]
@ -1306,7 +1306,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm2
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm0, %ymm3
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm4 = [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-NO-FMA-NEXT: vsubps %ymm3, %ymm4, %ymm3
; HASWELL-NO-FMA-NEXT: vmulps %ymm3, %ymm2, %ymm3
; HASWELL-NO-FMA-NEXT: vaddps %ymm3, %ymm2, %ymm2
@ -1328,7 +1328,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; KNL-LABEL: v16f32_two_step:
; KNL: # %bb.0:
; KNL-NEXT: vrcp14ps %zmm0, %zmm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [10:1.00]
; KNL-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] sched: [10:1.00]
; KNL-NEXT: vmovaps %zmm1, %zmm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} zmm3 = -(zmm0 * zmm3) + zmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} zmm3 = (zmm3 * zmm1) + zmm1 sched: [5:0.50]
@ -1339,7 +1339,7 @@ define <16 x float> @v16f32_two_step(<16 x float> %x) #2 {
; SKX-LABEL: v16f32_two_step:
; SKX: # %bb.0:
; SKX-NEXT: vrcp14ps %zmm0, %zmm1 # sched: [4:2.00]
; SKX-NEXT: vbroadcastss {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [8:0.50]
; SKX-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] sched: [8:0.50]
; SKX-NEXT: vmovaps %zmm1, %zmm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} zmm3 = -(zmm0 * zmm3) + zmm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} zmm3 = (zmm3 * zmm1) + zmm1 sched: [4:0.50]

162
test/CodeGen/X86/recip-fastmath2.ll
View File

@ -433,7 +433,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: subps %xmm0, %xmm1
; SSE-NEXT: mulps %xmm2, %xmm1
; SSE-NEXT: addps %xmm2, %xmm1
@ -445,7 +445,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX-RECIP-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: vaddps %xmm0, %xmm1, %xmm0
@ -470,7 +470,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
;
; BTVER2-LABEL: v4f32_one_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
@ -483,7 +483,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@ -493,7 +493,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; HASWELL-LABEL: v4f32_one_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; HASWELL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [11:0.50]
@ -503,7 +503,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@ -513,7 +513,7 @@ define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
; KNL-LABEL: v4f32_one_step2:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; KNL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [11:0.50]
@ -535,11 +535,11 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm1
; SSE-NEXT: mulps %xmm1, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: subps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm1, %xmm2
; SSE-NEXT: addps %xmm1, %xmm2
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,2,3,4]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; SSE-NEXT: mulps %xmm2, %xmm0
; SSE-NEXT: mulps %xmm2, %xmm0
; SSE-NEXT: retq
@ -548,7 +548,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %xmm0, %xmm2, %xmm0
; AVX-RECIP-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX-RECIP-NEXT: vaddps %xmm0, %xmm1, %xmm0
@ -576,7 +576,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
;
; BTVER2-LABEL: v4f32_one_step_2_divs:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
@ -590,7 +590,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@ -601,7 +601,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; HASWELL-LABEL: v4f32_one_step_2_divs:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; HASWELL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1 # sched: [11:0.50]
@ -612,7 +612,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm1, %xmm0, %xmm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm0, %xmm1, %xmm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
@ -623,7 +623,7 @@ define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
; KNL-LABEL: v4f32_one_step_2_divs:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + xmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1 sched: [5:0.50]
; KNL-NEXT: vmulps {{.*}}(%rip), %xmm0, %xmm1 # sched: [11:0.50]
@ -649,7 +649,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: movaps %xmm0, %xmm3
; SSE-NEXT: mulps %xmm2, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm1, %xmm4
; SSE-NEXT: subps %xmm3, %xmm4
; SSE-NEXT: mulps %xmm2, %xmm4
@ -666,7 +666,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %xmm0, %xmm1
; AVX-RECIP-NEXT: vmulps %xmm1, %xmm0, %xmm2
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %xmm2, %xmm3, %xmm2
; AVX-RECIP-NEXT: vmulps %xmm2, %xmm1, %xmm2
; AVX-RECIP-NEXT: vaddps %xmm2, %xmm1, %xmm1
@ -680,7 +680,7 @@ 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 = [1,1,1,1]
; 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
@ -692,7 +692,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; BDVER2-LABEL: v4f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:0.50]
; BDVER2-NEXT: vmovaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm3 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %xmm1, %xmm3, %xmm1, %xmm1 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %xmm2, %xmm1, %xmm0, %xmm0 # sched: [5:0.50]
@ -702,7 +702,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
;
; BTVER2-LABEL: v4f32_two_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %xmm0, %xmm1 # sched: [2:1.00]
; BTVER2-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [2:1.00]
; BTVER2-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
@ -719,7 +719,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; SANDY-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SANDY-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %xmm2, %xmm1, %xmm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
@ -733,7 +733,7 @@ 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 # sched: [5:1.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [5:0.50]
@ -746,7 +746,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm1, %xmm0, %xmm2 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %xmm2, %xmm1, %xmm2 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
@ -760,7 +760,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; KNL-LABEL: v4f32_two_step2:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %xmm0, %xmm1 # sched: [5:1.00]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; KNL-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; KNL-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [5:0.50]
@ -772,7 +772,7 @@ define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
; SKX-LABEL: v4f32_two_step2:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %xmm0, %xmm1 # sched: [4:1.00]
; SKX-NEXT: vbroadcastss {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
; SKX-NEXT: vbroadcastss {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [6:0.50]
; SKX-NEXT: vmovaps %xmm1, %xmm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1 sched: [4:0.50]
@ -789,7 +789,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm1, %xmm4
; SSE-NEXT: mulps %xmm4, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm2 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm2, %xmm3
; SSE-NEXT: subps %xmm1, %xmm3
; SSE-NEXT: mulps %xmm4, %xmm3
@ -809,7 +809,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm1, %ymm0
@ -834,7 +834,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
;
; BTVER2-LABEL: v8f32_one_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
@ -847,7 +847,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@ -857,7 +857,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; HASWELL-LABEL: v8f32_one_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; HASWELL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [12:0.50]
@ -867,7 +867,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-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] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@ -877,7 +877,7 @@ define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
; KNL-LABEL: v8f32_one_step2:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; KNL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [12:0.50]
@ -899,7 +899,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm3, %xmm4
; SSE-NEXT: subps %xmm0, %xmm4
; SSE-NEXT: mulps %xmm2, %xmm4
@ -909,9 +909,9 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; SSE-NEXT: subps %xmm1, %xmm3
; SSE-NEXT: mulps %xmm0, %xmm3
; SSE-NEXT: addps %xmm0, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm1 = [5,6,7,8]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; SSE-NEXT: mulps %xmm3, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,2,3,4]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; SSE-NEXT: mulps %xmm4, %xmm0
; SSE-NEXT: mulps %xmm4, %xmm0
; SSE-NEXT: mulps %xmm3, %xmm1
@ -921,7 +921,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm1, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm1, %ymm0
@ -949,7 +949,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
;
; BTVER2-LABEL: v8f32_one_step_2_divs:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
@ -963,7 +963,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@ -974,7 +974,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; HASWELL-LABEL: v8f32_one_step_2_divs:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; HASWELL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1 # sched: [12:0.50]
@ -985,7 +985,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm0, %ymm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-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] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm0, %ymm1, %ymm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
@ -996,7 +996,7 @@ define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
; KNL-LABEL: v8f32_one_step_2_divs:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + ymm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1 sched: [5:0.50]
; KNL-NEXT: vmulps {{.*}}(%rip), %ymm0, %ymm1 # sched: [12:0.50]
@ -1023,7 +1023,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; SSE-NEXT: rcpps %xmm1, %xmm3
; SSE-NEXT: movaps %xmm1, %xmm4
; SSE-NEXT: mulps %xmm3, %xmm4
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm0, %xmm5
; SSE-NEXT: subps %xmm4, %xmm5
; SSE-NEXT: mulps %xmm3, %xmm5
@ -1053,7 +1053,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm0, %ymm2
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm2, %ymm3, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm2
; AVX-RECIP-NEXT: vaddps %ymm2, %ymm1, %ymm1
@ -1067,7 +1067,7 @@ 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 = [1,1,1,1,1,1,1,1]
; 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
@ -1079,7 +1079,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; BDVER2-LABEL: v8f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm3 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %ymm1, %ymm3, %ymm1, %ymm1 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm2, %ymm1, %ymm0, %ymm0 # sched: [5:0.50]
@ -1089,7 +1089,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
;
; BTVER2-LABEL: v8f32_two_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm1 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:2.00]
@ -1106,7 +1106,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm1 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm2, %ymm1, %ymm2 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
@ -1120,7 +1120,7 @@ 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 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [5:0.50]
@ -1133,7 +1133,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm0, %ymm2 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-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] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm1, %ymm2 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
@ -1147,7 +1147,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; KNL-LABEL: v8f32_two_step2:
; KNL: # %bb.0:
; KNL-NEXT: vrcpps %ymm0, %ymm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; KNL-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [5:0.50]
@ -1159,7 +1159,7 @@ define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
; SKX-LABEL: v8f32_two_step2:
; SKX: # %bb.0:
; SKX-NEXT: vrcpps %ymm0, %ymm1 # sched: [4:1.00]
; SKX-NEXT: vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SKX-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] sched: [7:0.50]
; SKX-NEXT: vmovaps %ymm1, %ymm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1 sched: [4:0.50]
@ -1300,7 +1300,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; SSE-NEXT: movaps %xmm0, %xmm6
; SSE-NEXT: rcpps %xmm3, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm4
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm0, %xmm3
; SSE-NEXT: subps %xmm4, %xmm3
; SSE-NEXT: mulps %xmm2, %xmm3
@ -1333,7 +1333,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm1, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm1
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm1, %ymm3, %ymm1
; AVX-RECIP-NEXT: vmulps %ymm1, %ymm2, %ymm1
; AVX-RECIP-NEXT: vaddps %ymm1, %ymm2, %ymm1
@ -1349,7 +1349,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; FMA-RECIP-LABEL: v16f32_one_step2:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm1, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; 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 {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
@ -1362,7 +1362,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; BDVER2-LABEL: v16f32_one_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm1, %ymm2 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vrcpps %ymm0, %ymm4 # sched: [5:2.00]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm1 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm4, %ymm0, %ymm0 # sched: [5:0.50]
@ -1374,7 +1374,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
;
; BTVER2-LABEL: v16f32_one_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm1, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm4 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm1, %ymm1 # sched: [2:2.00]
@ -1393,7 +1393,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm1, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm1, %ymm1 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm1, %ymm3, %ymm1 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm1, %ymm2, %ymm1 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm1, %ymm2, %ymm1 # sched: [3:1.00]
@ -1409,7 +1409,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; HASWELL-LABEL: v16f32_one_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm1, %ymm2 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vrcpps %ymm0, %ymm4 # sched: [11:2.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2 sched: [5:0.50]
@ -1423,7 +1423,7 @@ define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm1, %ymm2 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm1, %ymm1 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-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] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm1, %ymm3, %ymm1 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm1, %ymm2, %ymm1 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %ymm1, %ymm2, %ymm1 # sched: [3:1.00]
@ -1460,7 +1460,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; SSE: # %bb.0:
; SSE-NEXT: rcpps %xmm0, %xmm6
; SSE-NEXT: mulps %xmm6, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm4 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm4, %xmm5
; SSE-NEXT: subps %xmm0, %xmm5
; SSE-NEXT: mulps %xmm6, %xmm5
@ -1482,13 +1482,13 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; SSE-NEXT: subps %xmm3, %xmm4
; SSE-NEXT: mulps %xmm0, %xmm4
; SSE-NEXT: addps %xmm0, %xmm4
; SSE-NEXT: movaps {{.*#+}} xmm3 = [13,14,15,16]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.3E+1,1.4E+1,1.5E+1,1.6E+1]
; SSE-NEXT: mulps %xmm4, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm2 = [9,10,11,12]
; SSE-NEXT: movaps {{.*#+}} xmm2 = [9.0E+0,1.0E+1,1.1E+1,1.2E+1]
; SSE-NEXT: mulps %xmm7, %xmm2
; SSE-NEXT: movaps {{.*#+}} xmm1 = [5,6,7,8]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [5.0E+0,6.0E+0,7.0E+0,8.0E+0]
; SSE-NEXT: mulps %xmm6, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,2,3,4]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; SSE-NEXT: mulps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm6, %xmm1
@ -1500,7 +1500,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm0, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm0, %ymm0
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX-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]
; AVX-RECIP-NEXT: vsubps %ymm0, %ymm3, %ymm0
; AVX-RECIP-NEXT: vmulps %ymm0, %ymm2, %ymm0
; AVX-RECIP-NEXT: vaddps %ymm0, %ymm2, %ymm0
@ -1518,7 +1518,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; FMA-RECIP-LABEL: v16f32_one_step_2_divs:
; FMA-RECIP: # %bb.0:
; FMA-RECIP-NEXT: vrcpps %ymm0, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; 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: vrcpps %ymm1, %ymm2
@ -1533,7 +1533,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; BDVER2-LABEL: v16f32_one_step_2_divs:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm0, %ymm0 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %ymm2, %ymm0, %ymm2, %ymm0 # sched: [5:0.50]
; BDVER2-NEXT: vrcpps %ymm1, %ymm2 # sched: [5:2.00]
@ -1547,7 +1547,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
;
; BTVER2-LABEL: v16f32_one_step_2_divs:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-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] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm0, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
@ -1568,7 +1568,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm0, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-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] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
; SANDY-NEXT: vrcpps %ymm1, %ymm4 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
@ -1586,7 +1586,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; HASWELL-LABEL: v16f32_one_step_2_divs:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm0, %ymm2 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2 sched: [5:0.50]
; HASWELL-NEXT: vrcpps %ymm1, %ymm2 # sched: [11:2.00]
@ -1602,7 +1602,7 @@ define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm0, %ymm2 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm0, %ymm0 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-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] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vrcpps %ymm1, %ymm4 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm0, %ymm2, %ymm0 # sched: [5:0.50]
@ -1647,7 +1647,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; SSE-NEXT: movaps %xmm0, %xmm4
; SSE-NEXT: rcpps %xmm3, %xmm2
; SSE-NEXT: mulps %xmm2, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm0, %xmm7
; SSE-NEXT: subps %xmm3, %xmm7
; SSE-NEXT: mulps %xmm2, %xmm7
@ -1703,7 +1703,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; AVX-RECIP: # %bb.0:
; AVX-RECIP-NEXT: vrcpps %ymm1, %ymm2
; AVX-RECIP-NEXT: vmulps %ymm2, %ymm1, %ymm3
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
; AVX-RECIP-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX-RECIP-NEXT: vsubps %ymm3, %ymm4, %ymm3
; AVX-RECIP-NEXT: vmulps %ymm3, %ymm2, %ymm3
; AVX-RECIP-NEXT: vaddps %ymm3, %ymm2, %ymm2
@ -1727,7 +1727,7 @@ 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 %ymm1, %ymm2
; FMA-RECIP-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; 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 = -(ymm1 * ymm4) + ymm3
; FMA-RECIP-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
@ -1746,7 +1746,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; BDVER2-LABEL: v16f32_two_step2:
; BDVER2: # %bb.0:
; BDVER2-NEXT: vrcpps %ymm1, %ymm2 # sched: [5:2.00]
; BDVER2-NEXT: vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:0.50]
; 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] sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm4 # sched: [5:0.50]
; BDVER2-NEXT: vfmaddps %ymm2, %ymm4, %ymm2, %ymm2 # sched: [5:0.50]
; BDVER2-NEXT: vfnmaddps %ymm3, %ymm2, %ymm1, %ymm1 # sched: [5:0.50]
@ -1762,7 +1762,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
;
; BTVER2-LABEL: v16f32_two_step2:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
; BTVER2-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [5:1.00]
; BTVER2-NEXT: vrcpps %ymm1, %ymm2 # sched: [2:2.00]
; BTVER2-NEXT: vmulps %ymm2, %ymm1, %ymm3 # sched: [2:2.00]
; BTVER2-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
@ -1789,7 +1789,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; SANDY: # %bb.0:
; SANDY-NEXT: vrcpps %ymm1, %ymm2 # sched: [7:2.00]
; SANDY-NEXT: vmulps %ymm2, %ymm1, %ymm3 # sched: [5:1.00]
; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; SANDY-NEXT: vmovaps {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; SANDY-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
; SANDY-NEXT: vmulps %ymm3, %ymm2, %ymm3 # sched: [5:1.00]
; SANDY-NEXT: vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
@ -1813,7 +1813,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; HASWELL-LABEL: v16f32_two_step2:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vrcpps %ymm1, %ymm2 # sched: [11:2.00]
; HASWELL-NEXT: vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; 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] sched: [7:0.50]
; HASWELL-NEXT: vmovaps %ymm2, %ymm4 # sched: [1:1.00]
; HASWELL-NEXT: vfnmadd213ps {{.*#+}} ymm4 = -(ymm1 * ymm4) + ymm3 sched: [5:0.50]
; HASWELL-NEXT: vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2 sched: [5:0.50]
@ -1833,7 +1833,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; HASWELL-NO-FMA: # %bb.0:
; HASWELL-NO-FMA-NEXT: vrcpps %ymm1, %ymm2 # sched: [11:2.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm2, %ymm1, %ymm3 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vbroadcastss {{.*#+}} ymm4 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0] sched: [7:0.50]
; HASWELL-NO-FMA-NEXT: vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
; HASWELL-NO-FMA-NEXT: vmulps %ymm3, %ymm2, %ymm3 # sched: [5:0.50]
; HASWELL-NO-FMA-NEXT: vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
@ -1857,7 +1857,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; KNL-LABEL: v16f32_two_step2:
; KNL: # %bb.0:
; KNL-NEXT: vrcp14ps %zmm0, %zmm1 # sched: [11:2.00]
; KNL-NEXT: vbroadcastss {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [10:1.00]
; KNL-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] sched: [10:1.00]
; KNL-NEXT: vmovaps %zmm1, %zmm3 # sched: [1:1.00]
; KNL-NEXT: vfnmadd213ps {{.*#+}} zmm3 = -(zmm0 * zmm3) + zmm2 sched: [5:0.50]
; KNL-NEXT: vfmadd132ps {{.*#+}} zmm3 = (zmm3 * zmm1) + zmm1 sched: [5:0.50]
@ -1869,7 +1869,7 @@ define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
; SKX-LABEL: v16f32_two_step2:
; SKX: # %bb.0:
; SKX-NEXT: vrcp14ps %zmm0, %zmm1 # sched: [4:2.00]
; SKX-NEXT: vbroadcastss {{.*#+}} zmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] sched: [8:0.50]
; SKX-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] sched: [8:0.50]
; SKX-NEXT: vmovaps %zmm1, %zmm3 # sched: [1:0.33]
; SKX-NEXT: vfnmadd213ps {{.*#+}} zmm3 = -(zmm0 * zmm3) + zmm2 sched: [4:0.50]
; SKX-NEXT: vfmadd132ps {{.*#+}} zmm3 = (zmm3 * zmm1) + zmm1 sched: [4:0.50]

4
test/CodeGen/X86/select_const.ll
View File

@ -468,10 +468,10 @@ define <2 x double> @sel_constants_fmul_constant_vec(i1 %cond) {
; CHECK-NEXT: testb $1, %dil
; CHECK-NEXT: jne .LBB37_1
; CHECK-NEXT: # %bb.2:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [118.83,34.539999999999999]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1.1883E+2,3.4539999999999999E+1]
; CHECK-NEXT: retq
; CHECK-NEXT: .LBB37_1:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [-20.399999999999999,37.68]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [-2.0399999999999999E+1,3.768E+1]
; CHECK-NEXT: retq
%sel = select i1 %cond, <2 x double> <double -4.0, double 12.0>, <2 x double> <double 23.3, double 11.0>
%bo = fmul <2 x double> %sel, <double 5.1, double 3.14>

8
test/CodeGen/X86/splat-for-size.ll
View File

@ -9,7 +9,7 @@
define <2 x double> @splat_v2f64(<2 x double> %x) #0 {
; CHECK-LABEL: splat_v2f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vmovddup {{.*#+}} xmm1 = [1,1]
; CHECK-NEXT: vmovddup {{.*#+}} xmm1 = [1.0E+0,1.0E+0]
; CHECK-NEXT: # xmm1 = mem[0,0]
; CHECK-NEXT: vaddpd %xmm1, %xmm0, %xmm0
; CHECK-NEXT: retq
@ -20,7 +20,7 @@ define <2 x double> @splat_v2f64(<2 x double> %x) #0 {
define <4 x double> @splat_v4f64(<4 x double> %x) #1 {
; CHECK-LABEL: splat_v4f64:
; CHECK: # %bb.0:
; CHECK-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1,1,1,1]
; CHECK-NEXT: vbroadcastsd {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; CHECK-NEXT: vaddpd %ymm1, %ymm0, %ymm0
; CHECK-NEXT: retq
%add = fadd <4 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0>
@ -30,7 +30,7 @@ define <4 x double> @splat_v4f64(<4 x double> %x) #1 {
define <4 x float> @splat_v4f32(<4 x float> %x) #0 {
; CHECK-LABEL: splat_v4f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; CHECK-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; CHECK-NEXT: vaddps %xmm1, %xmm0, %xmm0
; CHECK-NEXT: retq
%add = fadd <4 x float> %x, <float 1.0, float 1.0, float 1.0, float 1.0>
@ -40,7 +40,7 @@ define <4 x float> @splat_v4f32(<4 x float> %x) #0 {
define <8 x float> @splat_v8f32(<8 x float> %x) #1 {
; CHECK-LABEL: splat_v8f32:
; CHECK: # %bb.0:
; CHECK-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; CHECK-NEXT: vbroadcastss {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; CHECK-NEXT: vaddps %ymm1, %ymm0, %ymm0
; CHECK-NEXT: retq
%add = fadd <8 x float> %x, <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>

44
test/CodeGen/X86/sqrt-fastmath.ll
View File

@ -178,7 +178,7 @@ define <4 x float> @sqrt_v4f32_check_denorms(<4 x float> %x) #3 {
; SSE-NEXT: rsqrtps %xmm0, %xmm2
; SSE-NEXT: movaps %xmm0, %xmm1
; SSE-NEXT: mulps %xmm2, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm3 = [-0.5,-0.5,-0.5,-0.5]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; SSE-NEXT: mulps %xmm1, %xmm3
; SSE-NEXT: mulps %xmm2, %xmm1
; SSE-NEXT: addps {{.*}}(%rip), %xmm1
@ -208,9 +208,9 @@ define <4 x float> @sqrt_v4f32_check_denorms(<4 x float> %x) #3 {
; AVX512: # %bb.0:
; AVX512-NEXT: vrsqrtps %xmm0, %xmm1
; AVX512-NEXT: vmulps %xmm1, %xmm0, %xmm2
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm3 = [-3,-3,-3,-3]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm3 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; AVX512-NEXT: vfmadd231ps {{.*#+}} xmm3 = (xmm2 * xmm1) + xmm3
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [-0.5,-0.5,-0.5,-0.5]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; AVX512-NEXT: vmulps %xmm3, %xmm1, %xmm1
; AVX512-NEXT: vmulps %xmm1, %xmm2, %xmm1
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [NaN,NaN,NaN,NaN]
@ -282,21 +282,21 @@ define <4 x float> @v4f32_no_estimate(<4 x float> %x) #0 {
; SSE-LABEL: v4f32_no_estimate:
; SSE: # %bb.0:
; SSE-NEXT: sqrtps %xmm0, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: divps %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX1-LABEL: v4f32_no_estimate:
; AVX1: # %bb.0:
; AVX1-NEXT: vsqrtps %xmm0, %xmm0
; AVX1-NEXT: vmovaps {{.*#+}} xmm1 = [1,1,1,1]
; AVX1-NEXT: vmovaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX1-NEXT: vdivps %xmm0, %xmm1, %xmm0
; AVX1-NEXT: retq
;
; AVX512-LABEL: v4f32_no_estimate:
; AVX512: # %bb.0:
; AVX512-NEXT: vsqrtps %xmm0, %xmm0
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-NEXT: vdivps %xmm0, %xmm1, %xmm0
; AVX512-NEXT: retq
%sqrt = tail call <4 x float> @llvm.sqrt.v4f32(<4 x float> %x)
@ -331,9 +331,9 @@ define <4 x float> @v4f32_estimate(<4 x float> %x) #1 {
; AVX512: # %bb.0:
; AVX512-NEXT: vrsqrtps %xmm0, %xmm1
; AVX512-NEXT: vmulps %xmm1, %xmm0, %xmm0
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [-3,-3,-3,-3]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; AVX512-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + xmm2
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [-0.5,-0.5,-0.5,-0.5]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm2 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; AVX512-NEXT: vmulps %xmm0, %xmm2, %xmm0
; AVX512-NEXT: vmulps %xmm0, %xmm1, %xmm0
; AVX512-NEXT: retq
@ -347,7 +347,7 @@ define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
; SSE: # %bb.0:
; SSE-NEXT: sqrtps %xmm1, %xmm2
; SSE-NEXT: sqrtps %xmm0, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: divps %xmm3, %xmm0
; SSE-NEXT: divps %xmm2, %xmm1
@ -356,14 +356,14 @@ define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
; AVX1-LABEL: v8f32_no_estimate:
; AVX1: # %bb.0:
; AVX1-NEXT: vsqrtps %ymm0, %ymm0
; AVX1-NEXT: vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX1-NEXT: vmovaps {{.*#+}} ymm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX1-NEXT: vdivps %ymm0, %ymm1, %ymm0
; AVX1-NEXT: retq
;
; AVX512-LABEL: v8f32_no_estimate:
; AVX512: # %bb.0:
; AVX512-NEXT: vsqrtps %ymm0, %ymm0
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm1 = [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: vdivps %ymm0, %ymm1, %ymm0
; AVX512-NEXT: retq
%sqrt = tail call <8 x float> @llvm.sqrt.v8f32(<8 x float> %x)
@ -375,11 +375,11 @@ define <8 x float> @v8f32_estimate(<8 x float> %x) #1 {
; SSE-LABEL: v8f32_estimate:
; SSE: # %bb.0:
; SSE-NEXT: rsqrtps %xmm0, %xmm3
; SSE-NEXT: movaps {{.*#+}} xmm4 = [-0.5,-0.5,-0.5,-0.5]
; SSE-NEXT: movaps {{.*#+}} xmm4 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; SSE-NEXT: movaps %xmm3, %xmm2
; SSE-NEXT: mulps %xmm3, %xmm2
; SSE-NEXT: mulps %xmm0, %xmm2
; SSE-NEXT: movaps {{.*#+}} xmm0 = [-3,-3,-3,-3]
; SSE-NEXT: movaps {{.*#+}} xmm0 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; SSE-NEXT: addps %xmm0, %xmm2
; SSE-NEXT: mulps %xmm4, %xmm2
; SSE-NEXT: mulps %xmm3, %xmm2
@ -408,9 +408,9 @@ define <8 x float> @v8f32_estimate(<8 x float> %x) #1 {
; AVX512: # %bb.0:
; AVX512-NEXT: vrsqrtps %ymm0, %ymm1
; AVX512-NEXT: vmulps %ymm1, %ymm0, %ymm0
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [-3,-3,-3,-3,-3,-3,-3,-3]
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; AVX512-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + ymm2
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5]
; AVX512-NEXT: vbroadcastss {{.*#+}} ymm2 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; AVX512-NEXT: vmulps %ymm0, %ymm2, %ymm0
; AVX512-NEXT: vmulps %ymm0, %ymm1, %ymm0
; AVX512-NEXT: retq
@ -426,7 +426,7 @@ define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
; SSE-NEXT: sqrtps %xmm2, %xmm5
; SSE-NEXT: sqrtps %xmm1, %xmm2
; SSE-NEXT: sqrtps %xmm0, %xmm1
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1,1,1,1]
; SSE-NEXT: movaps {{.*#+}} xmm3 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movaps %xmm3, %xmm0
; SSE-NEXT: divps %xmm1, %xmm0
; SSE-NEXT: movaps %xmm3, %xmm1
@ -440,7 +440,7 @@ define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
; AVX1: # %bb.0:
; AVX1-NEXT: vsqrtps %ymm1, %ymm1
; AVX1-NEXT: vsqrtps %ymm0, %ymm0
; AVX1-NEXT: vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
; AVX1-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]
; AVX1-NEXT: vdivps %ymm0, %ymm2, %ymm0
; AVX1-NEXT: vdivps %ymm1, %ymm2, %ymm1
; AVX1-NEXT: retq
@ -448,7 +448,7 @@ define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
; AVX512-LABEL: v16f32_no_estimate:
; AVX512: # %bb.0:
; AVX512-NEXT: vsqrtps %zmm0, %zmm0
; AVX512-NEXT: vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
; AVX512-NEXT: vbroadcastss {{.*#+}} zmm1 = [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: vdivps %zmm0, %zmm1, %zmm0
; AVX512-NEXT: retq
%sqrt = tail call <16 x float> @llvm.sqrt.v16f32(<16 x float> %x)
@ -462,11 +462,11 @@ define <16 x float> @v16f32_estimate(<16 x float> %x) #1 {
; SSE-NEXT: movaps %xmm1, %xmm4
; SSE-NEXT: movaps %xmm0, %xmm1
; SSE-NEXT: rsqrtps %xmm0, %xmm5
; SSE-NEXT: movaps {{.*#+}} xmm6 = [-0.5,-0.5,-0.5,-0.5]
; SSE-NEXT: movaps {{.*#+}} xmm6 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; SSE-NEXT: movaps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm5, %xmm0
; SSE-NEXT: mulps %xmm1, %xmm0
; SSE-NEXT: movaps {{.*#+}} xmm7 = [-3,-3,-3,-3]
; SSE-NEXT: movaps {{.*#+}} xmm7 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; SSE-NEXT: addps %xmm7, %xmm0
; SSE-NEXT: mulps %xmm6, %xmm0
; SSE-NEXT: mulps %xmm5, %xmm0
@ -498,10 +498,10 @@ define <16 x float> @v16f32_estimate(<16 x float> %x) #1 {
; AVX1-LABEL: v16f32_estimate:
; AVX1: # %bb.0:
; AVX1-NEXT: vrsqrtps %ymm0, %ymm2
; AVX1-NEXT: vmovaps {{.*#+}} ymm3 = [-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5]
; AVX1-NEXT: vmovaps {{.*#+}} ymm3 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; AVX1-NEXT: vmulps %ymm2, %ymm2, %ymm4
; AVX1-NEXT: vmulps %ymm4, %ymm0, %ymm0
; AVX1-NEXT: vmovaps {{.*#+}} ymm4 = [-3,-3,-3,-3,-3,-3,-3,-3]
; AVX1-NEXT: vmovaps {{.*#+}} ymm4 = [-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0,-3.0E+0]
; AVX1-NEXT: vaddps %ymm4, %ymm0, %ymm0
; AVX1-NEXT: vmulps %ymm0, %ymm3, %ymm0
; AVX1-NEXT: vmulps %ymm0, %ymm2, %ymm0

4
test/CodeGen/X86/sse2.ll
View File

@ -395,7 +395,7 @@ define void @test12() nounwind {
; SSE-LABEL: test12:
; SSE: # %bb.0:
; SSE-NEXT: movapd 0, %xmm0
; SSE-NEXT: movapd {{.*#+}} xmm1 = [1,1,1,1]
; SSE-NEXT: movapd {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; SSE-NEXT: movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
; SSE-NEXT: xorps %xmm2, %xmm2
; SSE-NEXT: movhlps {{.*#+}} xmm2 = xmm0[1],xmm2[1]
@ -416,7 +416,7 @@ define void @test12() nounwind {
; AVX512-LABEL: test12:
; AVX512: # %bb.0:
; AVX512-NEXT: vmovaps 0, %xmm0
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512-NEXT: vbroadcastss {{.*#+}} xmm1 = [1.0E+0,1.0E+0,1.0E+0,1.0E+0]
; AVX512-NEXT: vblendps {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3]
; AVX512-NEXT: vxorps %xmm2, %xmm2, %xmm2
; AVX512-NEXT: vunpckhpd {{.*#+}} xmm0 = xmm0[1],xmm2[1]

8
test/CodeGen/X86/subvector-broadcast.ll
View File

@ -949,7 +949,7 @@ entry:
define void @fallback_broadcast_v4f64_to_v8f64(<4 x double> %a, <8 x double> %b) {
; X32-AVX-LABEL: fallback_broadcast_v4f64_to_v8f64:
; X32-AVX: # %bb.0: # %entry
; X32-AVX-NEXT: vmovapd {{.*#+}} ymm3 = [1,2,3,4]
; X32-AVX-NEXT: vmovapd {{.*#+}} ymm3 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; X32-AVX-NEXT: vaddpd %ymm3, %ymm0, %ymm0
; X32-AVX-NEXT: vaddpd %ymm3, %ymm2, %ymm2
; X32-AVX-NEXT: vaddpd %ymm3, %ymm1, %ymm1
@ -963,7 +963,7 @@ define void @fallback_broadcast_v4f64_to_v8f64(<4 x double> %a, <8 x double> %b)
;
; X32-AVX512-LABEL: fallback_broadcast_v4f64_to_v8f64:
; X32-AVX512: # %bb.0: # %entry
; X32-AVX512-NEXT: vmovapd {{.*#+}} ymm2 = [1,2,3,4]
; X32-AVX512-NEXT: vmovapd {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; X32-AVX512-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; X32-AVX512-NEXT: vinsertf64x4 $1, %ymm2, %zmm2, %zmm2
; X32-AVX512-NEXT: vaddpd %zmm2, %zmm1, %zmm1
@ -975,7 +975,7 @@ define void @fallback_broadcast_v4f64_to_v8f64(<4 x double> %a, <8 x double> %b)
;
; X64-AVX-LABEL: fallback_broadcast_v4f64_to_v8f64:
; X64-AVX: # %bb.0: # %entry
; X64-AVX-NEXT: vmovapd {{.*#+}} ymm3 = [1,2,3,4]
; X64-AVX-NEXT: vmovapd {{.*#+}} ymm3 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; X64-AVX-NEXT: vaddpd %ymm3, %ymm0, %ymm0
; X64-AVX-NEXT: vaddpd %ymm3, %ymm2, %ymm2
; X64-AVX-NEXT: vaddpd %ymm3, %ymm1, %ymm1
@ -989,7 +989,7 @@ define void @fallback_broadcast_v4f64_to_v8f64(<4 x double> %a, <8 x double> %b)
;
; X64-AVX512-LABEL: fallback_broadcast_v4f64_to_v8f64:
; X64-AVX512: # %bb.0: # %entry
; X64-AVX512-NEXT: vmovapd {{.*#+}} ymm2 = [1,2,3,4]
; X64-AVX512-NEXT: vmovapd {{.*#+}} ymm2 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; X64-AVX512-NEXT: vaddpd %ymm2, %ymm0, %ymm0
; X64-AVX512-NEXT: vinsertf64x4 $1, %ymm2, %zmm2, %zmm2
; X64-AVX512-NEXT: vaddpd %zmm2, %zmm1, %zmm1

4
test/CodeGen/X86/v4f32-immediate.ll
View File

@ -5,12 +5,12 @@
define <4 x float> @foo() {
; X32-LABEL: foo:
; X32: # %bb.0:
; X32-NEXT: movaps {{.*#+}} xmm0 = [3.22354245,2.29999995,1.20000005,0.100000001]
; X32-NEXT: movaps {{.*#+}} xmm0 = [3.22354245E+0,2.29999995E+0,1.20000005E+0,1.00000001E-1]
; X32-NEXT: retl
;
; X64-LABEL: foo:
; X64: # %bb.0:
; X64-NEXT: movaps {{.*#+}} xmm0 = [3.22354245,2.29999995,1.20000005,0.100000001]
; X64-NEXT: movaps {{.*#+}} xmm0 = [3.22354245E+0,2.29999995E+0,1.20000005E+0,1.00000001E-1]
; X64-NEXT: retq
ret <4 x float> <float 0x4009C9D0A0000000, float 0x4002666660000000, float 0x3FF3333340000000, float 0x3FB99999A0000000>
}

4
test/CodeGen/X86/vec_cast3.ll
View File

@ -90,7 +90,7 @@ define <2 x float> @cvt_v2u32_v2f32(<2 x i32> %src) {
; CHECK: ## %bb.0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; CHECK-NEXT: vmovaps {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
; CHECK-NEXT: vmovaps {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; CHECK-NEXT: vorps %xmm1, %xmm0, %xmm0
; CHECK-NEXT: vsubpd %xmm1, %xmm0, %xmm0
; CHECK-NEXT: vcvtpd2ps %xmm0, %xmm0
@ -99,7 +99,7 @@ define <2 x float> @cvt_v2u32_v2f32(<2 x i32> %src) {
; CHECK-WIDE-LABEL: cvt_v2u32_v2f32:
; CHECK-WIDE: ## %bb.0:
; CHECK-WIDE-NEXT: vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; CHECK-WIDE-NEXT: vmovdqa {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
; CHECK-WIDE-NEXT: vmovdqa {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; CHECK-WIDE-NEXT: vpor %xmm1, %xmm0, %xmm0
; CHECK-WIDE-NEXT: vsubpd %xmm1, %xmm0, %xmm0
; CHECK-WIDE-NEXT: vcvtpd2ps %xmm0, %xmm0

36
test/CodeGen/X86/vec_floor.ll
View File

@ -703,17 +703,17 @@ declare <16 x float> @llvm.nearbyint.v16f32(<16 x float> %p)
define <2 x double> @const_floor_v2f64() {
; SSE41-LABEL: const_floor_v2f64:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-2,2]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-2.0E+0,2.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_floor_v2f64:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-2,2]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-2.0E+0,2.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_floor_v2f64:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-2,2]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-2.0E+0,2.0E+0]
; AVX512-NEXT: retq
%t = call <2 x double> @llvm.floor.v2f64(<2 x double> <double -1.5, double 2.5>)
ret <2 x double> %t
@ -722,17 +722,17 @@ define <2 x double> @const_floor_v2f64() {
define <4 x float> @const_floor_v4f32() {
; SSE41-LABEL: const_floor_v4f32:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-4,6,-9,2]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-4.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_floor_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-4,6,-9,2]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-4.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_floor_v4f32:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-4,6,-9,2]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-4.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; AVX512-NEXT: retq
%t = call <4 x float> @llvm.floor.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
ret <4 x float> %t
@ -741,17 +741,17 @@ define <4 x float> @const_floor_v4f32() {
define <2 x double> @const_ceil_v2f64() {
; SSE41-LABEL: const_ceil_v2f64:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-1,3]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-1.0E+0,3.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_ceil_v2f64:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-1,3]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-1.0E+0,3.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_ceil_v2f64:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-1,3]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-1.0E+0,3.0E+0]
; AVX512-NEXT: retq
%t = call <2 x double> @llvm.ceil.v2f64(<2 x double> <double -1.5, double 2.5>)
ret <2 x double> %t
@ -760,17 +760,17 @@ define <2 x double> @const_ceil_v2f64() {
define <4 x float> @const_ceil_v4f32() {
; SSE41-LABEL: const_ceil_v4f32:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-3,6,-9,3]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,3.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_ceil_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-3,6,-9,3]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,3.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_ceil_v4f32:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-3,6,-9,3]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,3.0E+0]
; AVX512-NEXT: retq
%t = call <4 x float> @llvm.ceil.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
ret <4 x float> %t
@ -779,17 +779,17 @@ define <4 x float> @const_ceil_v4f32() {
define <2 x double> @const_trunc_v2f64() {
; SSE41-LABEL: const_trunc_v2f64:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-1,2]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-1.0E+0,2.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_trunc_v2f64:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-1,2]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-1.0E+0,2.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_trunc_v2f64:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-1,2]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-1.0E+0,2.0E+0]
; AVX512-NEXT: retq
%t = call <2 x double> @llvm.trunc.v2f64(<2 x double> <double -1.5, double 2.5>)
ret <2 x double> %t
@ -798,17 +798,17 @@ define <2 x double> @const_trunc_v2f64() {
define <4 x float> @const_trunc_v4f32() {
; SSE41-LABEL: const_trunc_v4f32:
; SSE41: ## %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-3,6,-9,2]
; SSE41-NEXT: movaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; SSE41-NEXT: retq
;
; AVX-LABEL: const_trunc_v4f32:
; AVX: ## %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-3,6,-9,2]
; AVX-NEXT: vmovaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; AVX-NEXT: retq
;
; AVX512-LABEL: const_trunc_v4f32:
; AVX512: ## %bb.0:
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-3,6,-9,2]
; AVX512-NEXT: vmovaps {{.*#+}} xmm0 = [-3.0E+0,6.0E+0,-9.0E+0,2.0E+0]
; AVX512-NEXT: retq
%t = call <4 x float> @llvm.trunc.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
ret <4 x float> %t

8
test/CodeGen/X86/vec_fp_to_int.ll
View File

@ -630,7 +630,7 @@ define <4 x i32> @fptoui_4f64_to_2i32(<2 x double> %a) {
; AVX1-LABEL: fptoui_4f64_to_2i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovapd %xmm0, %xmm0
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [2147483648,2147483648,2147483648,2147483648]
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [2.147483648E+9,2.147483648E+9,2.147483648E+9,2.147483648E+9]
; AVX1-NEXT: vcmpltpd %ymm1, %ymm0, %ymm2
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3
; AVX1-NEXT: vpackssdw %xmm3, %xmm2, %xmm2
@ -645,7 +645,7 @@ define <4 x i32> @fptoui_4f64_to_2i32(<2 x double> %a) {
; AVX2-LABEL: fptoui_4f64_to_2i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vmovapd %xmm0, %xmm0
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [2147483648,2147483648,2147483648,2147483648]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [2.147483648E+9,2.147483648E+9,2.147483648E+9,2.147483648E+9]
; AVX2-NEXT: vcmpltpd %ymm1, %ymm0, %ymm2
; AVX2-NEXT: vextractf128 $1, %ymm2, %xmm3
; AVX2-NEXT: vpackssdw %xmm3, %xmm2, %xmm2
@ -952,7 +952,7 @@ define <4 x i32> @fptoui_4f64_to_4i32(<4 x double> %a) {
;
; AVX1-LABEL: fptoui_4f64_to_4i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [2147483648,2147483648,2147483648,2147483648]
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [2.147483648E+9,2.147483648E+9,2.147483648E+9,2.147483648E+9]
; AVX1-NEXT: vcmpltpd %ymm1, %ymm0, %ymm2
; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3
; AVX1-NEXT: vpackssdw %xmm3, %xmm2, %xmm2
@ -966,7 +966,7 @@ define <4 x i32> @fptoui_4f64_to_4i32(<4 x double> %a) {
;
; AVX2-LABEL: fptoui_4f64_to_4i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [2147483648,2147483648,2147483648,2147483648]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [2.147483648E+9,2.147483648E+9,2.147483648E+9,2.147483648E+9]
; AVX2-NEXT: vcmpltpd %ymm1, %ymm0, %ymm2
; AVX2-NEXT: vextractf128 $1, %ymm2, %xmm3
; AVX2-NEXT: vpackssdw %xmm3, %xmm2, %xmm2

12
test/CodeGen/X86/vec_fpext.ll
View File

@ -253,42 +253,42 @@ entry:
define <2 x double> @fpext_fromconst() {
; X32-SSE-LABEL: fpext_fromconst:
; X32-SSE: # %bb.0: # %entry
; X32-SSE-NEXT: movaps {{.*#+}} xmm0 = [1,-2]
; X32-SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,-2.0E+0]
; X32-SSE-NEXT: # encoding: [0x0f,0x28,0x05,A,A,A,A]
; X32-SSE-NEXT: # fixup A - offset: 3, value: {{\.LCPI.*}}, kind: FK_Data_4
; X32-SSE-NEXT: retl # encoding: [0xc3]
;
; X32-AVX-LABEL: fpext_fromconst:
; X32-AVX: # %bb.0: # %entry
; X32-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1,-2]
; X32-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1.0E+0,-2.0E+0]
; X32-AVX-NEXT: # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; X32-AVX-NEXT: # fixup A - offset: 4, value: {{\.LCPI.*}}, kind: FK_Data_4
; X32-AVX-NEXT: retl # encoding: [0xc3]
;
; X32-AVX512VL-LABEL: fpext_fromconst:
; X32-AVX512VL: # %bb.0: # %entry
; X32-AVX512VL-NEXT: vmovaps {{\.LCPI.*}}, %xmm0 # EVEX TO VEX Compression xmm0 = [1,-2]
; X32-AVX512VL-NEXT: vmovaps {{\.LCPI.*}}, %xmm0 # EVEX TO VEX Compression xmm0 = [1.0E+0,-2.0E+0]
; X32-AVX512VL-NEXT: # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; X32-AVX512VL-NEXT: # fixup A - offset: 4, value: {{\.LCPI.*}}, kind: FK_Data_4
; X32-AVX512VL-NEXT: retl # encoding: [0xc3]
;
; X64-SSE-LABEL: fpext_fromconst:
; X64-SSE: # %bb.0: # %entry
; X64-SSE-NEXT: movaps {{.*#+}} xmm0 = [1,-2]
; X64-SSE-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,-2.0E+0]
; X64-SSE-NEXT: # encoding: [0x0f,0x28,0x05,A,A,A,A]
; X64-SSE-NEXT: # fixup A - offset: 3, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
; X64-SSE-NEXT: retq # encoding: [0xc3]
;
; X64-AVX-LABEL: fpext_fromconst:
; X64-AVX: # %bb.0: # %entry
; X64-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1,-2]
; X64-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1.0E+0,-2.0E+0]
; X64-AVX-NEXT: # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; X64-AVX-NEXT: # fixup A - offset: 4, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
; X64-AVX-NEXT: retq # encoding: [0xc3]
;
; X64-AVX512VL-LABEL: fpext_fromconst:
; X64-AVX512VL: # %bb.0: # %entry
; X64-AVX512VL-NEXT: vmovaps {{.*}}(%rip), %xmm0 # EVEX TO VEX Compression xmm0 = [1,-2]
; X64-AVX512VL-NEXT: vmovaps {{.*}}(%rip), %xmm0 # EVEX TO VEX Compression xmm0 = [1.0E+0,-2.0E+0]
; X64-AVX512VL-NEXT: # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
; X64-AVX512VL-NEXT: # fixup A - offset: 4, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
; X64-AVX512VL-NEXT: retq # encoding: [0xc3]

14
test/CodeGen/X86/vec_int_to_fp.ll
View File

@ -680,7 +680,7 @@ define <2 x double> @uitofp_4i32_to_2f64(<4 x i32> %a) {
; AVX2: # %bb.0:
; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
; AVX2-NEXT: vcvtdq2pd %xmm1, %ymm1
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [65536,65536,65536,65536]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [6.5536E+4,6.5536E+4,6.5536E+4,6.5536E+4]
; AVX2-NEXT: vmulpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
@ -954,7 +954,7 @@ define <4 x double> @uitofp_4i32_to_4f64(<4 x i32> %a) {
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: psrld $16, %xmm1
; SSE2-NEXT: cvtdq2pd %xmm1, %xmm1
; SSE2-NEXT: movapd {{.*#+}} xmm2 = [65536,65536]
; SSE2-NEXT: movapd {{.*#+}} xmm2 = [6.5536E+4,6.5536E+4]
; SSE2-NEXT: mulpd %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
; SSE2-NEXT: pand {{.*}}(%rip), %xmm0
@ -974,7 +974,7 @@ define <4 x double> @uitofp_4i32_to_4f64(<4 x i32> %a) {
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: psrld $16, %xmm1
; SSE41-NEXT: cvtdq2pd %xmm1, %xmm1
; SSE41-NEXT: movapd {{.*#+}} xmm2 = [65536,65536]
; SSE41-NEXT: movapd {{.*#+}} xmm2 = [6.5536E+4,6.5536E+4]
; SSE41-NEXT: mulpd %xmm2, %xmm1
; SSE41-NEXT: pxor %xmm3, %xmm3
; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
@ -1004,7 +1004,7 @@ define <4 x double> @uitofp_4i32_to_4f64(<4 x i32> %a) {
; AVX2: # %bb.0:
; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
; AVX2-NEXT: vcvtdq2pd %xmm1, %ymm1
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [65536,65536,65536,65536]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [6.5536E+4,6.5536E+4,6.5536E+4,6.5536E+4]
; AVX2-NEXT: vmulpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
@ -3777,7 +3777,7 @@ define <4 x double> @uitofp_load_4i32_to_4f64(<4 x i32> *%a) {
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: psrld $16, %xmm1
; SSE2-NEXT: cvtdq2pd %xmm1, %xmm1
; SSE2-NEXT: movapd {{.*#+}} xmm2 = [65536,65536]
; SSE2-NEXT: movapd {{.*#+}} xmm2 = [6.5536E+4,6.5536E+4]
; SSE2-NEXT: mulpd %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
; SSE2-NEXT: pand {{.*}}(%rip), %xmm0
@ -3798,7 +3798,7 @@ define <4 x double> @uitofp_load_4i32_to_4f64(<4 x i32> *%a) {
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: psrld $16, %xmm1
; SSE41-NEXT: cvtdq2pd %xmm1, %xmm1
; SSE41-NEXT: movapd {{.*#+}} xmm2 = [65536,65536]
; SSE41-NEXT: movapd {{.*#+}} xmm2 = [6.5536E+4,6.5536E+4]
; SSE41-NEXT: mulpd %xmm2, %xmm1
; SSE41-NEXT: pxor %xmm3, %xmm3
; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
@ -3830,7 +3830,7 @@ define <4 x double> @uitofp_load_4i32_to_4f64(<4 x i32> *%a) {
; AVX2-NEXT: vmovdqa (%rdi), %xmm0
; AVX2-NEXT: vpsrld $16, %xmm0, %xmm1
; AVX2-NEXT: vcvtdq2pd %xmm1, %ymm1
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [65536,65536,65536,65536]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [6.5536E+4,6.5536E+4,6.5536E+4,6.5536E+4]
; AVX2-NEXT: vmulpd %ymm2, %ymm1, %ymm1
; AVX2-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]

8
test/CodeGen/X86/vec_ss_load_fold.ll
View File

@ -247,22 +247,22 @@ define <4 x float> @test4(<4 x float> %A, float *%b, i32 %C) nounwind {
define <2 x double> @test5() nounwind uwtable readnone noinline {
; X32-LABEL: test5:
; X32: ## %bb.0: ## %entry
; X32-NEXT: movaps {{.*#+}} xmm0 = [128,123.321]
; X32-NEXT: movaps {{.*#+}} xmm0 = [1.28E+2,1.23321E+2]
; X32-NEXT: retl
;
; X64-LABEL: test5:
; X64: ## %bb.0: ## %entry
; X64-NEXT: movaps {{.*#+}} xmm0 = [128,123.321]
; X64-NEXT: movaps {{.*#+}} xmm0 = [1.28E+2,1.23321E+2]
; X64-NEXT: retq
;
; X32_AVX-LABEL: test5:
; X32_AVX: ## %bb.0: ## %entry
; X32_AVX-NEXT: vmovaps {{.*#+}} xmm0 = [128,123.321]
; X32_AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1.28E+2,1.23321E+2]
; X32_AVX-NEXT: retl
;
; X64_AVX-LABEL: test5:
; X64_AVX: ## %bb.0: ## %entry
; X64_AVX-NEXT: vmovaps {{.*#+}} xmm0 = [128,123.321]
; X64_AVX-NEXT: vmovaps {{.*#+}} xmm0 = [1.28E+2,1.23321E+2]
; X64_AVX-NEXT: retq
entry:
%0 = tail call <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double> <double 4.569870e+02, double 1.233210e+02>, i32 128) nounwind readnone

4
test/CodeGen/X86/vec_uint_to_fp-fastmath.ll
View File

@ -109,7 +109,7 @@ define <8 x float> @test_uitofp_v8i32_to_v8f32(<8 x i32> %arg) {
; SSE2-NEXT: movdqa %xmm0, %xmm2
; SSE2-NEXT: psrld $16, %xmm2
; SSE2-NEXT: cvtdq2ps %xmm2, %xmm2
; SSE2-NEXT: movaps {{.*#+}} xmm3 = [65536,65536,65536,65536]
; SSE2-NEXT: movaps {{.*#+}} xmm3 = [6.5536E+4,6.5536E+4,6.5536E+4,6.5536E+4]
; SSE2-NEXT: mulps %xmm3, %xmm2
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535]
; SSE2-NEXT: pand %xmm4, %xmm0
@ -129,7 +129,7 @@ define <8 x float> @test_uitofp_v8i32_to_v8f32(<8 x i32> %arg) {
; SSE41-NEXT: movdqa %xmm0, %xmm2
; SSE41-NEXT: psrld $16, %xmm2
; SSE41-NEXT: cvtdq2ps %xmm2, %xmm2
; SSE41-NEXT: movaps {{.*#+}} xmm3 = [65536,65536,65536,65536]
; SSE41-NEXT: movaps {{.*#+}} xmm3 = [6.5536E+4,6.5536E+4,6.5536E+4,6.5536E+4]
; SSE41-NEXT: mulps %xmm3, %xmm2
; SSE41-NEXT: pxor %xmm4, %xmm4
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3],xmm0[4],xmm4[5],xmm0[6],xmm4[7]

40
test/CodeGen/X86/vector-constrained-fp-intrinsics.ll
View File

@ -26,13 +26,13 @@ entry:
define <2 x double> @constrained_vector_fdiv_v2f64() {
; NO-FMA-LABEL: constrained_vector_fdiv_v2f64:
; NO-FMA: # %bb.0: # %entry
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1,2]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1.0E+0,2.0E+0]
; NO-FMA-NEXT: divpd {{.*}}(%rip), %xmm0
; NO-FMA-NEXT: retq
;
; HAS-FMA-LABEL: constrained_vector_fdiv_v2f64:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [1,2]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [1.0E+0,2.0E+0]
; HAS-FMA-NEXT: vdivpd {{.*}}(%rip), %xmm0, %xmm0
; HAS-FMA-NEXT: retq
entry:
@ -82,7 +82,7 @@ entry:
define <3 x double> @constrained_vector_fdiv_v3f64() {
; NO-FMA-LABEL: constrained_vector_fdiv_v3f64:
; NO-FMA: # %bb.0: # %entry
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1,2]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1.0E+0,2.0E+0]
; NO-FMA-NEXT: divpd {{.*}}(%rip), %xmm0
; NO-FMA-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
; NO-FMA-NEXT: divsd {{.*}}(%rip), %xmm1
@ -96,7 +96,7 @@ define <3 x double> @constrained_vector_fdiv_v3f64() {
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; HAS-FMA-NEXT: vdivsd {{.*}}(%rip), %xmm0, %xmm0
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm1 = [1,2]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm1 = [1.0E+0,2.0E+0]
; HAS-FMA-NEXT: vdivpd {{.*}}(%rip), %xmm1, %xmm1
; HAS-FMA-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; HAS-FMA-NEXT: retq
@ -112,16 +112,16 @@ entry:
define <4 x double> @constrained_vector_fdiv_v4f64() {
; NO-FMA-LABEL: constrained_vector_fdiv_v4f64:
; NO-FMA: # %bb.0: # %entry
; NO-FMA-NEXT: movapd {{.*#+}} xmm2 = [10,10]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1,2]
; NO-FMA-NEXT: movapd {{.*#+}} xmm2 = [1.0E+1,1.0E+1]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1.0E+0,2.0E+0]
; NO-FMA-NEXT: divpd %xmm2, %xmm0
; NO-FMA-NEXT: movapd {{.*#+}} xmm1 = [3,4]
; NO-FMA-NEXT: movapd {{.*#+}} xmm1 = [3.0E+0,4.0E+0]
; NO-FMA-NEXT: divpd %xmm2, %xmm1
; NO-FMA-NEXT: retq
;
; HAS-FMA-LABEL: constrained_vector_fdiv_v4f64:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm0 = [1,2,3,4]
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm0 = [1.0E+0,2.0E+0,3.0E+0,4.0E+0]
; HAS-FMA-NEXT: vdivpd {{.*}}(%rip), %ymm0, %ymm0
; HAS-FMA-NEXT: retq
entry:
@ -499,7 +499,7 @@ define <4 x double> @constrained_vector_fmul_v4f64() {
; NO-FMA-LABEL: constrained_vector_fmul_v4f64:
; NO-FMA: # %bb.0: # %entry
; NO-FMA-NEXT: movapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [2,3]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [2.0E+0,3.0E+0]
; NO-FMA-NEXT: mulpd %xmm1, %xmm0
; NO-FMA-NEXT: mulpd {{.*}}(%rip), %xmm1
; NO-FMA-NEXT: retq
@ -630,7 +630,7 @@ define <4 x double> @constrained_vector_fadd_v4f64() {
; NO-FMA-LABEL: constrained_vector_fadd_v4f64:
; NO-FMA: # %bb.0: # %entry
; NO-FMA-NEXT: movapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1,0.10000000000000001]
; NO-FMA-NEXT: movapd {{.*#+}} xmm0 = [1.0E+0,1.0000000000000001E-1]
; NO-FMA-NEXT: addpd %xmm1, %xmm0
; NO-FMA-NEXT: addpd {{.*}}(%rip), %xmm1
; NO-FMA-NEXT: retq
@ -836,8 +836,8 @@ define <2 x double> @constrained_vector_fma_v2f64() {
;
; HAS-FMA-LABEL: constrained_vector_fma_v2f64:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm1 = [1.5,0.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [3.5,2.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm1 = [1.5E+0,5.0E-1]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [3.5E+0,2.5E+0]
; HAS-FMA-NEXT: vfmadd213pd {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
; HAS-FMA-NEXT: retq
entry:
@ -936,8 +936,8 @@ define <3 x double> @constrained_vector_fma_v3f64() {
; HAS-FMA-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
; HAS-FMA-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
; HAS-FMA-NEXT: vfmadd213sd {{.*#+}} xmm1 = (xmm0 * xmm1) + mem
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [2.5,1.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm2 = [5.5,4.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm0 = [2.5E+0,1.5E+0]
; HAS-FMA-NEXT: vmovapd {{.*#+}} xmm2 = [5.5E+0,4.5E+0]
; HAS-FMA-NEXT: vfmadd213pd {{.*#+}} xmm2 = (xmm0 * xmm2) + mem
; HAS-FMA-NEXT: vinsertf128 $1, %xmm1, %ymm2, %ymm0
; HAS-FMA-NEXT: retq
@ -987,8 +987,8 @@ define <4 x double> @constrained_vector_fma_v4f64() {
;
; HAS-FMA-LABEL: constrained_vector_fma_v4f64:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm1 = [3.5,2.5,1.5,0.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm0 = [7.5,6.5,5.5,4.5]
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm1 = [3.5E+0,2.5E+0,1.5E+0,5.0E-1]
; HAS-FMA-NEXT: vmovapd {{.*#+}} ymm0 = [7.5E+0,6.5E+0,5.5E+0,4.5E+0]
; HAS-FMA-NEXT: vfmadd213pd {{.*#+}} ymm0 = (ymm1 * ymm0) + mem
; HAS-FMA-NEXT: retq
entry:
@ -1037,8 +1037,8 @@ define <4 x float> @constrained_vector_fma_v4f32() {
;
; HAS-FMA-LABEL: constrained_vector_fma_v4f32:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovaps {{.*#+}} xmm1 = [3.5,2.5,1.5,0.5]
; HAS-FMA-NEXT: vmovaps {{.*#+}} xmm0 = [7.5,6.5,5.5,4.5]
; HAS-FMA-NEXT: vmovaps {{.*#+}} xmm1 = [3.5E+0,2.5E+0,1.5E+0,5.0E-1]
; HAS-FMA-NEXT: vmovaps {{.*#+}} xmm0 = [7.5E+0,6.5E+0,5.5E+0,4.5E+0]
; HAS-FMA-NEXT: vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
; HAS-FMA-NEXT: retq
entry:
@ -1115,8 +1115,8 @@ define <8 x float> @constrained_vector_fma_v8f32() {
;
; HAS-FMA-LABEL: constrained_vector_fma_v8f32:
; HAS-FMA: # %bb.0: # %entry
; HAS-FMA-NEXT: vmovaps {{.*#+}} ymm1 = [3.5,2.5,1.5,0.5,7.5,6.5,5.5,4.5]
; HAS-FMA-NEXT: vmovaps {{.*#+}} ymm0 = [7.5,6.5,5.5,4.5,11.5,10.5,9.5,8.5]
; HAS-FMA-NEXT: vmovaps {{.*#+}} ymm1 = [3.5E+0,2.5E+0,1.5E+0,5.0E-1,7.5E+0,6.5E+0,5.5E+0,4.5E+0]
; HAS-FMA-NEXT: vmovaps {{.*#+}} ymm0 = [7.5E+0,6.5E+0,5.5E+0,4.5E+0,1.15E+1,1.05E+1,9.5E+0,8.5E+0]
; HAS-FMA-NEXT: vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + mem
; HAS-FMA-NEXT: retq
entry:

16
test/CodeGen/X86/vector-shuffle-combining-avx.ll
View File

@ -383,12 +383,12 @@ define <4 x float> @combine_vpermilvar_4f32_as_insertps(<4 x float> %a0) {
define <2 x double> @constant_fold_vpermilvar_pd() {
; X32-LABEL: constant_fold_vpermilvar_pd:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [2,1]
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [2.0E+0,1.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermilvar_pd:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [2,1]
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [2.0E+0,1.0E+0]
; X64-NEXT: retq
%1 = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> <double 1.0, double 2.0>, <2 x i64> <i64 2, i64 0>)
ret <2 x double> %1
@ -397,12 +397,12 @@ define <2 x double> @constant_fold_vpermilvar_pd() {
define <4 x double> @constant_fold_vpermilvar_pd_256() {
; X32-LABEL: constant_fold_vpermilvar_pd_256:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [2,1,3,4]
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [2.0E+0,1.0E+0,3.0E+0,4.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermilvar_pd_256:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [2,1,3,4]
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [2.0E+0,1.0E+0,3.0E+0,4.0E+0]
; X64-NEXT: retq
%1 = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> <double 1.0, double 2.0, double 3.0, double 4.0>, <4 x i64> <i64 2, i64 0, i64 0, i64 2>)
ret <4 x double> %1
@ -411,12 +411,12 @@ define <4 x double> @constant_fold_vpermilvar_pd_256() {
define <4 x float> @constant_fold_vpermilvar_ps() {
; X32-LABEL: constant_fold_vpermilvar_ps:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [4,1,3,2]
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [4.0E+0,1.0E+0,3.0E+0,2.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermilvar_ps:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [4,1,3,2]
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [4.0E+0,1.0E+0,3.0E+0,2.0E+0]
; X64-NEXT: retq
%1 = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, <4 x i32> <i32 3, i32 0, i32 2, i32 1>)
ret <4 x float> %1
@ -425,12 +425,12 @@ define <4 x float> @constant_fold_vpermilvar_ps() {
define <8 x float> @constant_fold_vpermilvar_ps_256() {
; X32-LABEL: constant_fold_vpermilvar_ps_256:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [1,1,3,2,5,6,6,6]
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [1.0E+0,1.0E+0,3.0E+0,2.0E+0,5.0E+0,6.0E+0,6.0E+0,6.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermilvar_ps_256:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [1,1,3,2,5,6,6,6]
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [1.0E+0,1.0E+0,3.0E+0,2.0E+0,5.0E+0,6.0E+0,6.0E+0,6.0E+0]
; X64-NEXT: retq
%1 = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<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>, <8 x i32> <i32 4, i32 0, i32 2, i32 1, i32 0, i32 1, i32 1, i32 1>)
ret <8 x float> %1

2
test/CodeGen/X86/vector-shuffle-combining-avx2.ll
View File

@ -691,7 +691,7 @@ define <8 x i32> @constant_fold_permd() {
define <8 x float> @constant_fold_permps() {
; CHECK-LABEL: constant_fold_permps:
; CHECK: # %bb.0:
; CHECK-NEXT: vmovaps {{.*#+}} ymm0 = [5,7,3,2,8,2,6,1]
; CHECK-NEXT: vmovaps {{.*#+}} ymm0 = [5.0E+0,7.0E+0,3.0E+0,2.0E+0,8.0E+0,2.0E+0,6.0E+0,1.0E+0]
; CHECK-NEXT: ret{{[l|q]}}
%1 = call <8 x float> @llvm.x86.avx2.permps(<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>, <8 x i32> <i32 4, i32 6, i32 2, i32 1, i32 7, i32 1, i32 5, i32 0>)
ret <8 x float> %1

16
test/CodeGen/X86/vector-shuffle-combining-xop.ll
View File

@ -384,12 +384,12 @@ define void @buildvector_v4f32_07z6(float %a, <4 x float> %b, <4 x float>* %ptr)
define <2 x double> @constant_fold_vpermil2pd() {
; X32-LABEL: constant_fold_vpermil2pd:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-2,2]
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-2.0E+0,2.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermil2pd:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-2,2]
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-2.0E+0,2.0E+0]
; X64-NEXT: retq
%1 = call <2 x double> @llvm.x86.xop.vpermil2pd(<2 x double> <double 1.0, double 2.0>, <2 x double> <double -2.0, double -1.0>, <2 x i64> <i64 4, i64 2>, i8 2)
ret <2 x double> %1
@ -398,12 +398,12 @@ define <2 x double> @constant_fold_vpermil2pd() {
define <4 x double> @constant_fold_vpermil2pd_256() {
; X32-LABEL: constant_fold_vpermil2pd_256:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [-4,0,4,3]
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [-4.0E+0,0.0E+0,4.0E+0,3.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermil2pd_256:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [-4,0,4,3]
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [-4.0E+0,0.0E+0,4.0E+0,3.0E+0]
; X64-NEXT: retq
%1 = call <4 x double> @llvm.x86.xop.vpermil2pd.256(<4 x double> <double 1.0, double 2.0, double 3.0, double 4.0>, <4 x double> <double -4.0, double -3.0, double -2.0, double -1.0>, <4 x i64> <i64 4, i64 8, i64 2, i64 0>, i8 2)
ret <4 x double> %1
@ -412,12 +412,12 @@ define <4 x double> @constant_fold_vpermil2pd_256() {
define <4 x float> @constant_fold_vpermil2ps() {
; X32-LABEL: constant_fold_vpermil2ps:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-4,1,3,0]
; X32-NEXT: vmovaps {{.*#+}} xmm0 = [-4.0E+0,1.0E+0,3.0E+0,0.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermil2ps:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-4,1,3,0]
; X64-NEXT: vmovaps {{.*#+}} xmm0 = [-4.0E+0,1.0E+0,3.0E+0,0.0E+0]
; X64-NEXT: retq
%1 = call <4 x float> @llvm.x86.xop.vpermil2ps(<4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, <4 x float> <float -4.0, float -3.0, float -2.0, float -1.0>, <4 x i32> <i32 4, i32 0, i32 2, i32 8>, i8 2)
ret <4 x float> %1
@ -426,12 +426,12 @@ define <4 x float> @constant_fold_vpermil2ps() {
define <8 x float> @constant_fold_vpermil2ps_256() {
; X32-LABEL: constant_fold_vpermil2ps_256:
; X32: # %bb.0:
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [-8,1,3,0,5,0,5,7]
; X32-NEXT: vmovaps {{.*#+}} ymm0 = [-8.0E+0,1.0E+0,3.0E+0,0.0E+0,5.0E+0,0.0E+0,5.0E+0,7.0E+0]
; X32-NEXT: retl
;
; X64-LABEL: constant_fold_vpermil2ps_256:
; X64: # %bb.0:
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [-8,1,3,0,5,0,5,7]
; X64-NEXT: vmovaps {{.*#+}} ymm0 = [-8.0E+0,1.0E+0,3.0E+0,0.0E+0,5.0E+0,0.0E+0,5.0E+0,7.0E+0]
; X64-NEXT: retq
%1 = call <8 x float> @llvm.x86.xop.vpermil2ps.256(<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>, <8 x float> <float -8.0, float -7.0, float -6.0, float -5.0, float -4.0, float -3.0, float -2.0, float -1.0>, <8 x i32> <i32 4, i32 0, i32 2, i32 8, i32 0, i32 8, i32 0, i32 2>, i8 2)
ret <8 x float> %1

8
test/CodeGen/X86/vector-shuffle-combining.ll
View File

@ -2642,14 +2642,14 @@ define void @combine_scalar_load_with_blend_with_zero(double* %a0, <4 x float>*
define <4 x float> @combine_constant_insertion_v4f32(float %f) {
; SSE2-LABEL: combine_constant_insertion_v4f32:
; SSE2: # %bb.0:
; SSE2-NEXT: movaps {{.*#+}} xmm1 = <u,4,5,3>
; SSE2-NEXT: movaps {{.*#+}} xmm1 = <u,4.0E+0,5.0E+0,3.0E+0>
; SSE2-NEXT: movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
; SSE2-NEXT: movaps %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: combine_constant_insertion_v4f32:
; SSSE3: # %bb.0:
; SSSE3-NEXT: movaps {{.*#+}} xmm1 = <u,4,5,3>
; SSSE3-NEXT: movaps {{.*#+}} xmm1 = <u,4.0E+0,5.0E+0,3.0E+0>
; SSSE3-NEXT: movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
; SSSE3-NEXT: movaps %xmm1, %xmm0
; SSSE3-NEXT: retq
@ -2809,14 +2809,14 @@ define <4 x float> @PR30264(<4 x float> %x) {
;
; SSE41-LABEL: PR30264:
; SSE41: # %bb.0:
; SSE41-NEXT: movaps {{.*#+}} xmm1 = <u,u,4,1>
; SSE41-NEXT: movaps {{.*#+}} xmm1 = <u,u,4.0E+0,1.0E+0>
; SSE41-NEXT: insertps {{.*#+}} xmm1 = xmm0[0],zero,xmm1[2,3]
; SSE41-NEXT: movaps %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: PR30264:
; AVX: # %bb.0:
; AVX-NEXT: vmovaps {{.*#+}} xmm1 = <u,u,4,1>
; AVX-NEXT: vmovaps {{.*#+}} xmm1 = <u,u,4.0E+0,1.0E+0>
; AVX-NEXT: vinsertps {{.*#+}} xmm0 = xmm0[0],zero,xmm1[2,3]
; AVX-NEXT: retq
%shuf1 = shufflevector <4 x float> %x, <4 x float> <float undef, float 0.0, float undef, float undef>, <4 x i32> <i32 0, i32 5, i32 undef, i32 undef>

6
test/CodeGen/X86/vselect-avx.ll
View File

@ -46,7 +46,7 @@ define void @test2(double** %call1559, i64 %indvars.iv4198, <4 x i1> %tmp1895) {
; AVX1-NEXT: vpmovsxdq %xmm0, %xmm0
; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT: movq (%rdi,%rsi,8), %rax
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [0.5,0.5,0.5,0.5]
; AVX1-NEXT: vmovapd {{.*#+}} ymm1 = [5.0E-1,5.0E-1,5.0E-1,5.0E-1]
; AVX1-NEXT: vblendvpd %ymm0, {{.*}}(%rip), %ymm1, %ymm0
; AVX1-NEXT: vmovupd %ymm0, (%rax)
; AVX1-NEXT: vzeroupper
@ -57,8 +57,8 @@ define void @test2(double** %call1559, i64 %indvars.iv4198, <4 x i1> %tmp1895) {
; AVX2-NEXT: vpslld $31, %xmm0, %xmm0
; AVX2-NEXT: vpmovsxdq %xmm0, %ymm0
; AVX2-NEXT: movq (%rdi,%rsi,8), %rax
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [-0.5,-0.5,-0.5,-0.5]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [0.5,0.5,0.5,0.5]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm1 = [-5.0E-1,-5.0E-1,-5.0E-1,-5.0E-1]
; AVX2-NEXT: vbroadcastsd {{.*#+}} ymm2 = [5.0E-1,5.0E-1,5.0E-1,5.0E-1]
; AVX2-NEXT: vblendvpd %ymm0, %ymm1, %ymm2, %ymm0
; AVX2-NEXT: vmovupd %ymm0, (%rax)
; AVX2-NEXT: vzeroupper

2
test/CodeGen/X86/widen_arith-6.ll
View File

@ -14,7 +14,7 @@ define void @update(<3 x float>* %dst, <3 x float>* %src, i32 %n) nounwind {
; CHECK-NEXT: movl $1073741824, {{[0-9]+}}(%esp) # imm = 0x40000000
; CHECK-NEXT: movl $1065353216, {{[0-9]+}}(%esp) # imm = 0x3F800000
; CHECK-NEXT: movl $0, {{[0-9]+}}(%esp)
; CHECK-NEXT: movaps {{.*#+}} xmm0 = <1976.04004,1976.04004,1976.04004,u>
; CHECK-NEXT: movaps {{.*#+}} xmm0 = <1.97604004E+3,1.97604004E+3,1.97604004E+3,u>
; CHECK-NEXT: jmp .LBB0_1
; CHECK-NEXT: .p2align 4, 0x90
; CHECK-NEXT: .LBB0_2: # %forbody

4
test/CodeGen/X86/x86-setcc-int-to-fp-combine.ll
View File

@ -52,7 +52,7 @@ define void @foo2(<4 x float>* noalias %result) nounwind {
; CHECK-NEXT: .long 1088421888 ## float 7
; CHECK-LABEL: foo2:
; CHECK: ## %bb.0:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [4,5,6,7]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [4.0E+0,5.0E+0,6.0E+0,7.0E+0]
; CHECK-NEXT: movaps %xmm0, (%rdi)
; CHECK-NEXT: retq
%val = uitofp <4 x i32> <i32 4, i32 5, i32 6, i32 7> to <4 x float>
@ -89,7 +89,7 @@ define void @foo4(<4 x float>* noalias %result) nounwind {
; CHECK-NEXT: .long 1132396544 ## float 255
; CHECK-LABEL: foo4:
; CHECK: ## %bb.0:
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1,127,128,255]
; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1.0E+0,1.27E+2,1.28E+2,2.55E+2]
; CHECK-NEXT: movaps %xmm0, (%rdi)
; CHECK-NEXT: retq
%val = uitofp <4 x i8> <i8 1, i8 127, i8 -128, i8 -1> to <4 x float>