RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-20 03:23:01 +02:00
Code Issues Projects Releases Wiki Activity

[X86] Make AVX512_512_SET0 XMM16-31 lower to 128-bit XOR when AVX512VL is enabled. Use 128-bit VLX instruction when VLX is enabled.

Browse Source 
Unfortunately, this weakens our ability to do domain fixing when AVX512DQ is not enabled, but it is consistent with our 256-bit behavior.

Maybe we should add custom handling to domain fixing to allow EVEX integer XOR/AND/OR/ANDN to switch to VEX encoded fp instructions if the high registers aren't being used?

llvm-svn: 316978
This commit is contained in:
Craig Topper 2017-10-31 06:01:04 +00:00
parent ffcc4669b7
commit 232eac1fb9
10 changed files with 320 additions and 331 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
lib/Target/X86/X86InstrInfo.cpp
View File

@ -7852,7 +7852,8 @@ bool X86InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
MIB->getOperand(0).setReg(SrcReg);
return Expand2AddrUndef(MIB, get(X86::VPXORDZrr));
}
case X86::AVX512_256_SET0: {
case X86::AVX512_256_SET0:
case X86::AVX512_512_SET0: {
bool HasVLX = Subtarget.hasVLX();
unsigned SrcReg = MIB->getOperand(0).getReg();
const TargetRegisterInfo *TRI = &getRegisterInfo();
@ -7866,18 +7867,6 @@ bool X86InstrInfo::expandPostRAPseudo(MachineInstr &MI) const {
}
return Expand2AddrUndef(MIB, get(X86::VPXORDZrr));
}
case X86::AVX512_512_SET0: {
const TargetRegisterInfo *TRI = &getRegisterInfo();
unsigned SrcReg = MIB->getOperand(0).getReg();
if (TRI->getEncodingValue(SrcReg) < 16) {
unsigned XReg = TRI->getSubReg(SrcReg, X86::sub_xmm);
MIB->getOperand(0).setReg(XReg);
Expand2AddrUndef(MIB, get(X86::VXORPSrr));
MIB.addReg(SrcReg, RegState::ImplicitDefine);
return true;
}
return Expand2AddrUndef(MIB, get(X86::VPXORDZrr));
}
case X86::V_SETALLONES:
return Expand2AddrUndef(MIB, get(HasAVX ? X86::VPCMPEQDrr : X86::PCMPEQDrr));
case X86::AVX2_SETALLONES:

4
test/CodeGen/X86/avx512-cvt.ll
View File

@ -1468,7 +1468,7 @@ define <16 x double> @sbto16f64(<16 x double> %a) {
;
; VLNODQ-LABEL: sbto16f64:
; VLNODQ: # BB#0:
; VLNODQ-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; VLNODQ-NEXT: vpxor %xmm2, %xmm2, %xmm2
; VLNODQ-NEXT: vcmpltpd %zmm1, %zmm2, %k1
; VLNODQ-NEXT: vcmpltpd %zmm0, %zmm2, %k2
; VLNODQ-NEXT: vpcmpeqd %ymm1, %ymm1, %ymm1
@ -1513,7 +1513,7 @@ define <8 x double> @sbto8f64(<8 x double> %a) {
;
; VLNODQ-LABEL: sbto8f64:
; VLNODQ: # BB#0:
; VLNODQ-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; VLNODQ-NEXT: vpxor %xmm1, %xmm1, %xmm1
; VLNODQ-NEXT: vcmpltpd %zmm0, %zmm1, %k1
; VLNODQ-NEXT: vpcmpeqd %ymm0, %ymm0, %ymm0
; VLNODQ-NEXT: vmovdqa32 %ymm0, %ymm0 {%k1} {z}

64
test/CodeGen/X86/avx512-shuffles/broadcast-scalar-fp.ll
View File

@ -134,7 +134,7 @@ define <8 x double> @test_double_to_8(double %s) {
define <8 x double> @test_masked_double_to_8_mask0(double %s, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -149,7 +149,7 @@ define <8 x double> @test_masked_double_to_8_mask0(double %s, <8 x double> %defa
define <8 x double> @test_masked_z_double_to_8_mask0(double %s, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -162,7 +162,7 @@ define <8 x double> @test_masked_z_double_to_8_mask0(double %s, <8 x double> %ma
define <8 x double> @test_masked_double_to_8_mask1(double %s, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -177,7 +177,7 @@ define <8 x double> @test_masked_double_to_8_mask1(double %s, <8 x double> %defa
define <8 x double> @test_masked_z_double_to_8_mask1(double %s, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -190,7 +190,7 @@ define <8 x double> @test_masked_z_double_to_8_mask1(double %s, <8 x double> %ma
define <8 x double> @test_masked_double_to_8_mask2(double %s, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -205,7 +205,7 @@ define <8 x double> @test_masked_double_to_8_mask2(double %s, <8 x double> %defa
define <8 x double> @test_masked_z_double_to_8_mask2(double %s, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -218,7 +218,7 @@ define <8 x double> @test_masked_z_double_to_8_mask2(double %s, <8 x double> %ma
define <8 x double> @test_masked_double_to_8_mask3(double %s, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -233,7 +233,7 @@ define <8 x double> @test_masked_double_to_8_mask3(double %s, <8 x double> %defa
define <8 x double> @test_masked_z_double_to_8_mask3(double %s, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -497,7 +497,7 @@ define <16 x float> @test_float_to_16(float %s) {
define <16 x float> @test_masked_float_to_16_mask0(float %s, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -512,7 +512,7 @@ define <16 x float> @test_masked_float_to_16_mask0(float %s, <16 x float> %defau
define <16 x float> @test_masked_z_float_to_16_mask0(float %s, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -525,7 +525,7 @@ define <16 x float> @test_masked_z_float_to_16_mask0(float %s, <16 x float> %mas
define <16 x float> @test_masked_float_to_16_mask1(float %s, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -540,7 +540,7 @@ define <16 x float> @test_masked_float_to_16_mask1(float %s, <16 x float> %defau
define <16 x float> @test_masked_z_float_to_16_mask1(float %s, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -553,7 +553,7 @@ define <16 x float> @test_masked_z_float_to_16_mask1(float %s, <16 x float> %mas
define <16 x float> @test_masked_float_to_16_mask2(float %s, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -568,7 +568,7 @@ define <16 x float> @test_masked_float_to_16_mask2(float %s, <16 x float> %defau
define <16 x float> @test_masked_z_float_to_16_mask2(float %s, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -581,7 +581,7 @@ define <16 x float> @test_masked_z_float_to_16_mask2(float %s, <16 x float> %mas
define <16 x float> @test_masked_float_to_16_mask3(float %s, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -596,7 +596,7 @@ define <16 x float> @test_masked_float_to_16_mask3(float %s, <16 x float> %defau
define <16 x float> @test_masked_z_float_to_16_mask3(float %s, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -745,7 +745,7 @@ define <8 x double> @test_double_to_8_mem(double* %p) {
define <8 x double> @test_masked_double_to_8_mem_mask0(double* %p, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -760,7 +760,7 @@ define <8 x double> @test_masked_double_to_8_mem_mask0(double* %p, <8 x double>
define <8 x double> @test_masked_z_double_to_8_mem_mask0(double* %p, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -774,7 +774,7 @@ define <8 x double> @test_masked_z_double_to_8_mem_mask0(double* %p, <8 x double
define <8 x double> @test_masked_double_to_8_mem_mask1(double* %p, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -789,7 +789,7 @@ define <8 x double> @test_masked_double_to_8_mem_mask1(double* %p, <8 x double>
define <8 x double> @test_masked_z_double_to_8_mem_mask1(double* %p, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -803,7 +803,7 @@ define <8 x double> @test_masked_z_double_to_8_mem_mask1(double* %p, <8 x double
define <8 x double> @test_masked_double_to_8_mem_mask2(double* %p, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -818,7 +818,7 @@ define <8 x double> @test_masked_double_to_8_mem_mask2(double* %p, <8 x double>
define <8 x double> @test_masked_z_double_to_8_mem_mask2(double* %p, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -832,7 +832,7 @@ define <8 x double> @test_masked_z_double_to_8_mem_mask2(double* %p, <8 x double
define <8 x double> @test_masked_double_to_8_mem_mask3(double* %p, <8 x double> %default, <8 x double> %mask) {
; CHECK-LABEL: test_masked_double_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -847,7 +847,7 @@ define <8 x double> @test_masked_double_to_8_mem_mask3(double* %p, <8 x double>
define <8 x double> @test_masked_z_double_to_8_mem_mask3(double* %p, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_double_to_8_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -1123,7 +1123,7 @@ define <16 x float> @test_float_to_16_mem(float* %p) {
define <16 x float> @test_masked_float_to_16_mem_mask0(float* %p, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -1138,7 +1138,7 @@ define <16 x float> @test_masked_float_to_16_mem_mask0(float* %p, <16 x float> %
define <16 x float> @test_masked_z_float_to_16_mem_mask0(float* %p, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -1152,7 +1152,7 @@ define <16 x float> @test_masked_z_float_to_16_mem_mask0(float* %p, <16 x float>
define <16 x float> @test_masked_float_to_16_mem_mask1(float* %p, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -1167,7 +1167,7 @@ define <16 x float> @test_masked_float_to_16_mem_mask1(float* %p, <16 x float> %
define <16 x float> @test_masked_z_float_to_16_mem_mask1(float* %p, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -1181,7 +1181,7 @@ define <16 x float> @test_masked_z_float_to_16_mem_mask1(float* %p, <16 x float>
define <16 x float> @test_masked_float_to_16_mem_mask2(float* %p, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -1196,7 +1196,7 @@ define <16 x float> @test_masked_float_to_16_mem_mask2(float* %p, <16 x float> %
define <16 x float> @test_masked_z_float_to_16_mem_mask2(float* %p, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -1210,7 +1210,7 @@ define <16 x float> @test_masked_z_float_to_16_mem_mask2(float* %p, <16 x float>
define <16 x float> @test_masked_float_to_16_mem_mask3(float* %p, <16 x float> %default, <16 x float> %mask) {
; CHECK-LABEL: test_masked_float_to_16_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
; CHECK-NEXT: retq
@ -1225,7 +1225,7 @@ define <16 x float> @test_masked_float_to_16_mem_mask3(float* %p, <16 x float> %
define <16 x float> @test_masked_z_float_to_16_mem_mask3(float* %p, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_float_to_16_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
; CHECK-NEXT: retq

40
test/CodeGen/X86/avx512-shuffles/duplicate-high.ll
View File

@ -576,7 +576,7 @@ define <16 x float> @test_16xfloat_dup_high(<16 x float> %vec) {
define <16 x float> @test_masked_16xfloat_dup_high_mask0(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -590,7 +590,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mask0(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_dup_high_mask0(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -602,7 +602,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mask0(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_dup_high_mask1(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -616,7 +616,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mask1(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_dup_high_mask1(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -628,7 +628,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mask1(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_dup_high_mask2(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -642,7 +642,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mask2(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_dup_high_mask2(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -654,7 +654,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mask2(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_dup_high_mask3(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -668,7 +668,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mask3(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_dup_high_mask3(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -680,7 +680,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mask3(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_dup_high_mask4(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -694,7 +694,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mask4(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_dup_high_mask4(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -715,7 +715,7 @@ define <16 x float> @test_16xfloat_dup_high_mem(<16 x float>* %vp) {
define <16 x float> @test_masked_16xfloat_dup_high_mem_mask0(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -729,7 +729,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mem_mask0(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask0(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -742,7 +742,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask0(<16 x float>* %vp
define <16 x float> @test_masked_16xfloat_dup_high_mem_mask1(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -756,7 +756,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mem_mask1(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask1(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -769,7 +769,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask1(<16 x float>* %vp
define <16 x float> @test_masked_16xfloat_dup_high_mem_mask2(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -783,7 +783,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mem_mask2(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask2(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -796,7 +796,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask2(<16 x float>* %vp
define <16 x float> @test_masked_16xfloat_dup_high_mem_mask3(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -810,7 +810,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mem_mask3(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask3(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -823,7 +823,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask3(<16 x float>* %vp
define <16 x float> @test_masked_16xfloat_dup_high_mem_mask4(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq
@ -837,7 +837,7 @@ define <16 x float> @test_masked_16xfloat_dup_high_mem_mask4(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_dup_high_mem_mask4(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_high_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15]
; CHECK-NEXT: retq

80
test/CodeGen/X86/avx512-shuffles/duplicate-low.ll
View File

@ -417,7 +417,7 @@ define <8 x double> @test_8xdouble_dup_low(<8 x double> %vec) {
define <8 x double> @test_masked_8xdouble_dup_low_mask0(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -431,7 +431,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mask0(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_dup_low_mask0(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -443,7 +443,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mask0(<8 x double> %vec, <8
define <8 x double> @test_masked_8xdouble_dup_low_mask1(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -457,7 +457,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mask1(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_dup_low_mask1(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -469,7 +469,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mask1(<8 x double> %vec, <8
define <8 x double> @test_masked_8xdouble_dup_low_mask2(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -483,7 +483,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mask2(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_dup_low_mask2(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -495,7 +495,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mask2(<8 x double> %vec, <8
define <8 x double> @test_masked_8xdouble_dup_low_mask3(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -509,7 +509,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mask3(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_dup_low_mask3(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -521,7 +521,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mask3(<8 x double> %vec, <8
define <8 x double> @test_masked_8xdouble_dup_low_mask4(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -535,7 +535,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mask4(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_dup_low_mask4(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -556,7 +556,7 @@ define <8 x double> @test_8xdouble_dup_low_mem(<8 x double>* %vp) {
define <8 x double> @test_masked_8xdouble_dup_low_mem_mask0(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -570,7 +570,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mem_mask0(<8 x double>* %vp, <
define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask0(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -583,7 +583,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask0(<8 x double>* %vp,
define <8 x double> @test_masked_8xdouble_dup_low_mem_mask1(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -597,7 +597,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mem_mask1(<8 x double>* %vp, <
define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask1(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -610,7 +610,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask1(<8 x double>* %vp,
define <8 x double> @test_masked_8xdouble_dup_low_mem_mask2(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -624,7 +624,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mem_mask2(<8 x double>* %vp, <
define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask2(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -637,7 +637,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask2(<8 x double>* %vp,
define <8 x double> @test_masked_8xdouble_dup_low_mem_mask3(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -651,7 +651,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mem_mask3(<8 x double>* %vp, <
define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask3(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -664,7 +664,7 @@ define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask3(<8 x double>* %vp,
define <8 x double> @test_masked_8xdouble_dup_low_mem_mask4(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -678,7 +678,7 @@ define <8 x double> @test_masked_8xdouble_dup_low_mem_mask4(<8 x double>* %vp, <
define <8 x double> @test_masked_z_8xdouble_dup_low_mem_mask4(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6]
; CHECK-NEXT: retq
@ -1263,7 +1263,7 @@ define <16 x float> @test_16xfloat_dup_low(<16 x float> %vec) {
define <16 x float> @test_masked_16xfloat_dup_low_mask0(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -1277,7 +1277,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mask0(<16 x float> %vec, <16 x
define <16 x float> @test_masked_z_16xfloat_dup_low_mask0(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1289,7 +1289,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mask0(<16 x float> %vec, <16
define <16 x float> @test_masked_16xfloat_dup_low_mask1(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -1303,7 +1303,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mask1(<16 x float> %vec, <16 x
define <16 x float> @test_masked_z_16xfloat_dup_low_mask1(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1315,7 +1315,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mask1(<16 x float> %vec, <16
define <16 x float> @test_masked_16xfloat_dup_low_mask2(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -1329,7 +1329,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mask2(<16 x float> %vec, <16 x
define <16 x float> @test_masked_z_16xfloat_dup_low_mask2(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1341,7 +1341,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mask2(<16 x float> %vec, <16
define <16 x float> @test_masked_16xfloat_dup_low_mask3(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -1355,7 +1355,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mask3(<16 x float> %vec, <16 x
define <16 x float> @test_masked_z_16xfloat_dup_low_mask3(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1367,7 +1367,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mask3(<16 x float> %vec, <16
define <16 x float> @test_masked_16xfloat_dup_low_mask4(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -1381,7 +1381,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mask4(<16 x float> %vec, <16 x
define <16 x float> @test_masked_z_16xfloat_dup_low_mask4(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1402,7 +1402,7 @@ define <16 x float> @test_16xfloat_dup_low_mem(<16 x float>* %vp) {
define <16 x float> @test_masked_16xfloat_dup_low_mem_mask0(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1416,7 +1416,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mem_mask0(<16 x float>* %vp, <
define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask0(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1429,7 +1429,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask0(<16 x float>* %vp,
define <16 x float> @test_masked_16xfloat_dup_low_mem_mask1(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1443,7 +1443,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mem_mask1(<16 x float>* %vp, <
define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask1(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1456,7 +1456,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask1(<16 x float>* %vp,
define <16 x float> @test_masked_16xfloat_dup_low_mem_mask2(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1470,7 +1470,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mem_mask2(<16 x float>* %vp, <
define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask2(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1483,7 +1483,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask2(<16 x float>* %vp,
define <16 x float> @test_masked_16xfloat_dup_low_mem_mask3(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1497,7 +1497,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mem_mask3(<16 x float>* %vp, <
define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask3(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1510,7 +1510,7 @@ define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask3(<16 x float>* %vp,
define <16 x float> @test_masked_16xfloat_dup_low_mem_mask4(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq
@ -1524,7 +1524,7 @@ define <16 x float> @test_masked_16xfloat_dup_low_mem_mask4(<16 x float>* %vp, <
define <16 x float> @test_masked_z_16xfloat_dup_low_mem_mask4(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_dup_low_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 {%k1} {z} = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14]
; CHECK-NEXT: retq

96
test/CodeGen/X86/avx512-shuffles/in_lane_permute.ll
View File

@ -757,7 +757,7 @@ define <16 x float> @test_16xfloat_perm_mask0(<16 x float> %vec) {
define <16 x float> @test_masked_16xfloat_perm_mask0(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,1,3,1,6,4,6,5,8,9,8,11,13,13,13,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -771,7 +771,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask0(<16 x float> %vec, <16 x fl
define <16 x float> @test_masked_z_16xfloat_perm_mask0(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,3,1,6,4,6,5,8,9,8,11,13,13,13,15]
; CHECK-NEXT: retq
@ -783,7 +783,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask0(<16 x float> %vec, <16 x
define <16 x float> @test_masked_16xfloat_perm_imm_mask1(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[2,2,2,1,6,6,6,5,10,10,10,9,14,14,14,13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -797,7 +797,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mask1(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_perm_imm_mask1(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,2,2,1,6,6,6,5,10,10,10,9,14,14,14,13]
; CHECK-NEXT: retq
@ -809,7 +809,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_imm_mask1(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_perm_mask2(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,2,0,0,5,4,6,5,11,10,9,9,14,13,14,12]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -823,7 +823,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask2(<16 x float> %vec, <16 x fl
define <16 x float> @test_masked_z_16xfloat_perm_mask2(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,0,0,5,4,6,5,11,10,9,9,14,13,14,12]
; CHECK-NEXT: retq
@ -843,7 +843,7 @@ define <16 x float> @test_16xfloat_perm_imm_mask3(<16 x float> %vec) {
define <16 x float> @test_masked_16xfloat_perm_imm_mask3(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,1,0,2,5,5,4,6,9,9,8,10,13,13,12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -857,7 +857,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mask3(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_perm_imm_mask3(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,1,0,2,5,5,4,6,9,9,8,10,13,13,12,14]
; CHECK-NEXT: retq
@ -869,7 +869,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_imm_mask3(<16 x float> %vec, <1
define <16 x float> @test_masked_16xfloat_perm_mask4(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,2,3,3,5,5,5,7,11,11,8,11,14,12,14,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -883,7 +883,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask4(<16 x float> %vec, <16 x fl
define <16 x float> @test_masked_z_16xfloat_perm_mask4(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,3,3,5,5,5,7,11,11,8,11,14,12,14,15]
; CHECK-NEXT: retq
@ -895,7 +895,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask4(<16 x float> %vec, <16 x
define <16 x float> @test_masked_16xfloat_perm_imm_mask5(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[1,2,1,0,5,6,5,4,9,10,9,8,13,14,13,12]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -909,7 +909,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mask5(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_perm_imm_mask5(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2,1,0,5,6,5,4,9,10,9,8,13,14,13,12]
; CHECK-NEXT: retq
@ -929,7 +929,7 @@ define <16 x float> @test_16xfloat_perm_mask6(<16 x float> %vec) {
define <16 x float> @test_masked_16xfloat_perm_mask6(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[2,0,3,2,4,4,6,7,9,11,8,11,13,12,13,13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -943,7 +943,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask6(<16 x float> %vec, <16 x fl
define <16 x float> @test_masked_z_16xfloat_perm_mask6(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,0,3,2,4,4,6,7,9,11,8,11,13,12,13,13]
; CHECK-NEXT: retq
@ -955,7 +955,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask6(<16 x float> %vec, <16 x
define <16 x float> @test_masked_16xfloat_perm_imm_mask7(<16 x float> %vec, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm1 {%k1} = zmm0[3,3,0,2,7,7,4,6,11,11,8,10,15,15,12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -969,7 +969,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mask7(<16 x float> %vec, <16
define <16 x float> @test_masked_z_16xfloat_perm_imm_mask7(<16 x float> %vec, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,3,0,2,7,7,4,6,11,11,8,10,15,15,12,14]
; CHECK-NEXT: retq
@ -992,7 +992,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask0(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm2
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = zmm2[3,3,3,0,6,6,6,6,11,10,9,10,12,14,12,12]
; CHECK-NEXT: retq
@ -1007,7 +1007,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask0(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm1
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm1[3,3,3,0,6,6,6,6,11,10,9,10,12,14,12,12]
; CHECK-NEXT: retq
@ -1021,7 +1021,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask0(<16 x float>* %vp, <1
define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask1(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,3,2,1,5,7,6,5,9,11,10,9,13,15,14,13]
; CHECK-NEXT: retq
@ -1035,7 +1035,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask1(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_perm_imm_mem_mask1(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,3,2,1,5,7,6,5,9,11,10,9,13,15,14,13]
; CHECK-NEXT: retq
@ -1050,7 +1050,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask2(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm2
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = zmm2[2,0,0,3,5,5,6,5,9,8,8,8,14,12,13,13]
; CHECK-NEXT: retq
@ -1065,7 +1065,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask2(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm1
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm1[2,0,0,3,5,5,6,5,9,8,8,8,14,12,13,13]
; CHECK-NEXT: retq
@ -1088,7 +1088,7 @@ define <16 x float> @test_16xfloat_perm_imm_mem_mask3(<16 x float>* %vp) {
define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask3(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,0,3,1,5,4,7,5,9,8,11,9,13,12,15,13]
; CHECK-NEXT: retq
@ -1102,7 +1102,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask3(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_perm_imm_mem_mask3(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,0,3,1,5,4,7,5,9,8,11,9,13,12,15,13]
; CHECK-NEXT: retq
@ -1117,7 +1117,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask4(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm2
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = zmm2[3,3,1,1,6,5,5,6,11,11,10,9,15,14,12,12]
; CHECK-NEXT: retq
@ -1132,7 +1132,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask4(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm1
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm1[3,3,1,1,6,5,5,6,11,11,10,9,15,14,12,12]
; CHECK-NEXT: retq
@ -1146,7 +1146,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask4(<16 x float>* %vp, <1
define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask5(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[2,0,0,1,6,4,4,5,10,8,8,9,14,12,12,13]
; CHECK-NEXT: retq
@ -1160,7 +1160,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask5(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_perm_imm_mem_mask5(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[2,0,0,1,6,4,4,5,10,8,8,9,14,12,12,13]
; CHECK-NEXT: retq
@ -1185,7 +1185,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask6(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm2
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = zmm2[2,1,1,2,6,5,5,7,9,11,9,9,12,15,14,15]
; CHECK-NEXT: retq
@ -1200,7 +1200,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask6(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps (%rdi), %zmm1
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm1[2,1,1,2,6,5,5,7,9,11,9,9,12,15,14,15]
; CHECK-NEXT: retq
@ -1214,7 +1214,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask6(<16 x float>* %vp, <1
define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask7(<16 x float>* %vp, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_masked_16xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} = mem[1,2,0,1,5,6,4,5,9,10,8,9,13,14,12,13]
; CHECK-NEXT: retq
@ -1228,7 +1228,7 @@ define <16 x float> @test_masked_16xfloat_perm_imm_mem_mask7(<16 x float>* %vp,
define <16 x float> @test_masked_z_16xfloat_perm_imm_mem_mask7(<16 x float>* %vp, <16 x float> %mask) {
; CHECK-LABEL: test_masked_z_16xfloat_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqps %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[1,2,0,1,5,6,4,5,9,10,8,9,13,14,12,13]
; CHECK-NEXT: retq
@ -1625,7 +1625,7 @@ define <8 x double> @test_8xdouble_perm_mask0(<8 x double> %vec) {
define <8 x double> @test_masked_8xdouble_perm_mask0(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,0,3,2,4,5,7,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1639,7 +1639,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask0(<8 x double> %vec, <8 x dou
define <8 x double> @test_masked_z_8xdouble_perm_mask0(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,3,2,4,5,7,6]
; CHECK-NEXT: retq
@ -1651,7 +1651,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask0(<8 x double> %vec, <8 x d
define <8 x double> @test_masked_8xdouble_perm_mask1(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3,4,4,7,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1665,7 +1665,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask1(<8 x double> %vec, <8 x dou
define <8 x double> @test_masked_z_8xdouble_perm_mask1(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,4,4,7,6]
; CHECK-NEXT: retq
@ -1677,7 +1677,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask1(<8 x double> %vec, <8 x d
define <8 x double> @test_masked_8xdouble_perm_mask2(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,0,2,3,5,5,6,7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1691,7 +1691,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask2(<8 x double> %vec, <8 x dou
define <8 x double> @test_masked_z_8xdouble_perm_mask2(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,3,5,5,6,7]
; CHECK-NEXT: retq
@ -1711,7 +1711,7 @@ define <8 x double> @test_8xdouble_perm_mask3(<8 x double> %vec) {
define <8 x double> @test_masked_8xdouble_perm_mask3(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,2,4,4,6,7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1725,7 +1725,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask3(<8 x double> %vec, <8 x dou
define <8 x double> @test_masked_z_8xdouble_perm_mask3(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,2,4,4,6,7]
; CHECK-NEXT: retq
@ -1746,7 +1746,7 @@ define <8 x double> @test_8xdouble_perm_mem_mask0(<8 x double>* %vp) {
define <8 x double> @test_masked_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[0,1,2,3,5,4,7,6]
; CHECK-NEXT: retq
@ -1760,7 +1760,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[0,1,2,3,5,4,7,6]
; CHECK-NEXT: retq
@ -1774,7 +1774,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8
define <8 x double> @test_masked_8xdouble_perm_mem_mask1(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[0,1,3,3,4,5,7,7]
; CHECK-NEXT: retq
@ -1788,7 +1788,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask1(<8 x double>* %vp, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_mem_mask1(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[0,1,3,3,4,5,7,7]
; CHECK-NEXT: retq
@ -1802,7 +1802,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask1(<8 x double>* %vp, <8
define <8 x double> @test_masked_8xdouble_perm_mem_mask2(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[1,1,3,3,5,4,7,6]
; CHECK-NEXT: retq
@ -1816,7 +1816,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask2(<8 x double>* %vp, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_mem_mask2(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[1,1,3,3,5,4,7,6]
; CHECK-NEXT: retq
@ -1839,7 +1839,7 @@ define <8 x double> @test_8xdouble_perm_mem_mask3(<8 x double>* %vp) {
define <8 x double> @test_masked_8xdouble_perm_mem_mask3(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} = mem[1,0,3,2,4,5,6,7]
; CHECK-NEXT: retq
@ -1853,7 +1853,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask3(<8 x double>* %vp, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_mem_mask3(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermilpd {{.*#+}} zmm0 {%k1} {z} = mem[1,0,3,2,4,5,6,7]
; CHECK-NEXT: retq

96
test/CodeGen/X86/avx512-shuffles/permute.ll
View File

@ -2117,7 +2117,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask0(<16 x float> %vec, <16 x fl
; CHECK-LABEL: test_masked_16xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm3 = [15,7,5,13,4,9,11,13,12,6,0,0,11,15,5,7]
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2132,7 +2132,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask0(<16 x float> %vec, <16 x
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [15,7,5,13,4,9,11,13,12,6,0,0,11,15,5,7]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2145,7 +2145,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask1(<16 x float> %vec, <16 x fl
; CHECK-LABEL: test_masked_16xfloat_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm3 = [11,10,4,10,4,5,8,11,2,0,10,0,0,3,10,1]
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2160,7 +2160,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask1(<16 x float> %vec, <16 x
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [11,10,4,10,4,5,8,11,2,0,10,0,0,3,10,1]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2173,7 +2173,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask2(<16 x float> %vec, <16 x fl
; CHECK-LABEL: test_masked_16xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm3 = [0,15,6,14,3,6,5,2,5,15,11,6,6,4,8,11]
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2188,7 +2188,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask2(<16 x float> %vec, <16 x
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [0,15,6,14,3,6,5,2,5,15,11,6,6,4,8,11]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2210,7 +2210,7 @@ define <16 x float> @test_masked_16xfloat_perm_mask3(<16 x float> %vec, <16 x fl
; CHECK-LABEL: test_masked_16xfloat_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm3 = [10,7,0,14,6,6,0,2,13,8,11,2,5,13,13,3]
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2225,7 +2225,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mask3(<16 x float> %vec, <16 x
; CHECK-LABEL: test_masked_z_16xfloat_perm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [10,7,0,14,6,6,0,2,13,8,11,2,5,13,13,3]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2248,7 +2248,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask0(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [10,2,1,14,9,9,7,2,9,4,12,11,0,14,0,1]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2263,7 +2263,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask0(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [10,2,1,14,9,9,7,2,9,4,12,11,0,14,0,1]
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2278,7 +2278,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask1(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [4,2,3,5,11,6,4,7,6,4,14,8,15,12,9,4]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2293,7 +2293,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask1(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [4,2,3,5,11,6,4,7,6,4,14,8,15,12,9,4]
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2308,7 +2308,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask2(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [10,7,11,6,7,0,11,0,10,9,12,4,10,3,8,5]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2323,7 +2323,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask2(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [10,7,11,6,7,0,11,0,10,9,12,4,10,3,8,5]
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2348,7 +2348,7 @@ define <16 x float> @test_masked_16xfloat_perm_mem_mask3(<16 x float>* %vp, <16
; CHECK-LABEL: test_masked_16xfloat_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm2 = [15,15,3,9,5,15,14,9,11,10,5,14,14,5,11,0]
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2363,7 +2363,7 @@ define <16 x float> @test_masked_z_16xfloat_perm_mem_mask3(<16 x float>* %vp, <1
; CHECK-LABEL: test_masked_z_16xfloat_perm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vmovaps {{.*#+}} zmm1 = [15,15,3,9,5,15,14,9,11,10,5,14,14,5,11,0]
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermps (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2637,7 +2637,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask0(<8 x double> %vec, <8 x dou
; CHECK-LABEL: test_masked_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm3 = [5,7,4,2,7,4,3,4]
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2652,7 +2652,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask0(<8 x double> %vec, <8 x d
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [5,7,4,2,7,4,3,4]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2664,7 +2664,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask0(<8 x double> %vec, <8 x d
define <8 x double> @test_masked_8xdouble_perm_imm_mask1(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm1 {%k1} = zmm0[3,0,0,2,7,4,4,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2678,7 +2678,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mask1(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_imm_mask1(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = zmm0[3,0,0,2,7,4,4,6]
; CHECK-NEXT: retq
@ -2691,7 +2691,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask2(<8 x double> %vec, <8 x dou
; CHECK-LABEL: test_masked_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm3 = [7,5,5,5,3,5,1,7]
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2706,7 +2706,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask2(<8 x double> %vec, <8 x d
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [7,5,5,5,3,5,1,7]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2726,7 +2726,7 @@ define <8 x double> @test_8xdouble_perm_imm_mask3(<8 x double> %vec) {
define <8 x double> @test_masked_8xdouble_perm_imm_mask3(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm1 {%k1} = zmm0[1,3,3,0,5,7,7,4]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2740,7 +2740,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mask3(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_imm_mask3(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1,3,3,0,5,7,7,4]
; CHECK-NEXT: retq
@ -2753,7 +2753,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask4(<8 x double> %vec, <8 x dou
; CHECK-LABEL: test_masked_8xdouble_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm3 = [3,5,3,4,6,5,7,1]
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2768,7 +2768,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask4(<8 x double> %vec, <8 x d
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [3,5,3,4,6,5,7,1]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2780,7 +2780,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask4(<8 x double> %vec, <8 x d
define <8 x double> @test_masked_8xdouble_perm_imm_mask5(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm1 {%k1} = zmm0[3,3,2,3,7,7,6,7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2794,7 +2794,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mask5(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_imm_mask5(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = zmm0[3,3,2,3,7,7,6,7]
; CHECK-NEXT: retq
@ -2816,7 +2816,7 @@ define <8 x double> @test_masked_8xdouble_perm_mask6(<8 x double> %vec, <8 x dou
; CHECK-LABEL: test_masked_8xdouble_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm3 = [2,7,6,4,0,0,0,2]
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm2, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm3, %zmm1 {%k1}
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2831,7 +2831,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask6(<8 x double> %vec, <8 x d
; CHECK-LABEL: test_masked_z_8xdouble_perm_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [2,7,6,4,0,0,0,2]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd %zmm0, %zmm2, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2843,7 +2843,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mask6(<8 x double> %vec, <8 x d
define <8 x double> @test_masked_8xdouble_perm_imm_mask7(<8 x double> %vec, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm1 {%k1} = zmm0[3,1,3,2,7,5,7,6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2857,7 +2857,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mask7(<8 x double> %vec, <8 x
define <8 x double> @test_masked_z_8xdouble_perm_imm_mask7(<8 x double> %vec, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = zmm0[3,1,3,2,7,5,7,6]
; CHECK-NEXT: retq
@ -2880,7 +2880,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8 x
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [0,3,4,0,4,2,0,1]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2895,7 +2895,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm1 = [0,3,4,0,4,2,0,1]
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2909,7 +2909,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask0(<8 x double>* %vp, <8
define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask1(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} = mem[0,2,0,3,4,6,4,7]
; CHECK-NEXT: retq
@ -2923,7 +2923,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask1(<8 x double>* %vp,
define <8 x double> @test_masked_z_8xdouble_perm_imm_mem_mask1(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = mem[0,2,0,3,4,6,4,7]
; CHECK-NEXT: retq
@ -2938,7 +2938,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask2(<8 x double>* %vp, <8 x
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [6,7,2,7,7,6,2,5]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -2953,7 +2953,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask2(<8 x double>* %vp, <8
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm1 = [6,7,2,7,7,6,2,5]
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -2976,7 +2976,7 @@ define <8 x double> @test_8xdouble_perm_imm_mem_mask3(<8 x double>* %vp) {
define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask3(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} = mem[2,1,1,0,6,5,5,4]
; CHECK-NEXT: retq
@ -2990,7 +2990,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask3(<8 x double>* %vp,
define <8 x double> @test_masked_z_8xdouble_perm_imm_mem_mask3(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = mem[2,1,1,0,6,5,5,4]
; CHECK-NEXT: retq
@ -3005,7 +3005,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask4(<8 x double>* %vp, <8 x
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [1,1,3,5,6,0,6,0]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -3020,7 +3020,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask4(<8 x double>* %vp, <8
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask4:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm1 = [1,1,3,5,6,0,6,0]
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -3034,7 +3034,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask4(<8 x double>* %vp, <8
define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask5(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} = mem[2,2,2,3,6,6,6,7]
; CHECK-NEXT: retq
@ -3048,7 +3048,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask5(<8 x double>* %vp,
define <8 x double> @test_masked_z_8xdouble_perm_imm_mem_mask5(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mem_mask5:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = mem[2,2,2,3,6,6,6,7]
; CHECK-NEXT: retq
@ -3073,7 +3073,7 @@ define <8 x double> @test_masked_8xdouble_perm_mem_mask6(<8 x double>* %vp, <8 x
; CHECK-LABEL: test_masked_8xdouble_perm_mem_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm2 = [2,4,0,4,6,1,2,5]
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm1, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm2, %zmm0 {%k1}
; CHECK-NEXT: retq
@ -3088,7 +3088,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask6(<8 x double>* %vp, <8
; CHECK-LABEL: test_masked_z_8xdouble_perm_mem_mask6:
; CHECK: # BB#0:
; CHECK-NEXT: vmovapd {{.*#+}} zmm1 = [2,4,0,4,6,1,2,5]
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm0, %k1
; CHECK-NEXT: vpermpd (%rdi), %zmm1, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
@ -3102,7 +3102,7 @@ define <8 x double> @test_masked_z_8xdouble_perm_mem_mask6(<8 x double>* %vp, <8
define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask7(<8 x double>* %vp, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_masked_8xdouble_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} = mem[0,3,2,0,4,7,6,4]
; CHECK-NEXT: retq
@ -3116,7 +3116,7 @@ define <8 x double> @test_masked_8xdouble_perm_imm_mem_mask7(<8 x double>* %vp,
define <8 x double> @test_masked_z_8xdouble_perm_imm_mem_mask7(<8 x double>* %vp, <8 x double> %mask) {
; CHECK-LABEL: test_masked_z_8xdouble_perm_imm_mem_mask7:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1
; CHECK-NEXT: vcmpeqpd %zmm1, %zmm0, %k1
; CHECK-NEXT: vpermpd {{.*#+}} zmm0 {%k1} {z} = mem[0,3,2,0,4,7,6,4]
; CHECK-NEXT: retq

64
test/CodeGen/X86/avx512-shuffles/shuffle-interleave.ll
View File

@ -520,7 +520,7 @@ define <16 x float> @test_16xfloat_shuff_mask0(<16 x float> %vec1, <16 x float>
define <16 x float> @test_16xfloat_masked_shuff_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[3,2],zmm1[3,2],zmm0[7,6],zmm1[7,6],zmm0[11,10],zmm1[11,10],zmm0[15,14],zmm1[15,14]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -534,7 +534,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask0(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,2],zmm1[3,2],zmm0[7,6],zmm1[7,6],zmm0[11,10],zmm1[11,10],zmm0[15,14],zmm1[15,14]
; CHECK-NEXT: retq
@ -546,7 +546,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mask0(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_masked_shuff_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[1,2],zmm1[3,3],zmm0[5,6],zmm1[7,7],zmm0[9,10],zmm1[11,11],zmm0[13,14],zmm1[15,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -560,7 +560,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask1(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[1,2],zmm1[3,3],zmm0[5,6],zmm1[7,7],zmm0[9,10],zmm1[11,11],zmm0[13,14],zmm1[15,15]
; CHECK-NEXT: retq
@ -572,7 +572,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mask1(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_masked_shuff_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[3,0],zmm1[2,1],zmm0[7,4],zmm1[6,5],zmm0[11,8],zmm1[10,9],zmm0[15,12],zmm1[14,13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -586,7 +586,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask2(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,0],zmm1[2,1],zmm0[7,4],zmm1[6,5],zmm0[11,8],zmm1[10,9],zmm0[15,12],zmm1[14,13]
; CHECK-NEXT: retq
@ -606,7 +606,7 @@ define <16 x float> @test_16xfloat_shuff_mask3(<16 x float> %vec1, <16 x float>
define <16 x float> @test_16xfloat_masked_shuff_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm2 {%k1} = zmm0[2,3],zmm1[0,2],zmm0[6,7],zmm1[4,6],zmm0[10,11],zmm1[8,10],zmm0[14,15],zmm1[12,14]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -620,7 +620,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask3(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3],zmm1[0,2],zmm0[6,7],zmm1[4,6],zmm0[10,11],zmm1[8,10],zmm0[14,15],zmm1[12,14]
; CHECK-NEXT: retq
@ -641,7 +641,7 @@ define <16 x float> @test_16xfloat_shuff_mem_mask0(<16 x float> %vec1, <16 x flo
define <16 x float> @test_16xfloat_masked_shuff_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[3,0],mem[0,2],zmm0[7,4],mem[4,6],zmm0[11,8],mem[8,10],zmm0[15,12],mem[12,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -656,7 +656,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask0(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,0],mem[0,2],zmm0[7,4],mem[4,6],zmm0[11,8],mem[8,10],zmm0[15,12],mem[12,14]
; CHECK-NEXT: retq
@ -670,7 +670,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask0(<16 x float> %vec
define <16 x float> @test_16xfloat_masked_shuff_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[0,2],mem[3,2],zmm0[4,6],mem[7,6],zmm0[8,10],mem[11,10],zmm0[12,14],mem[15,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -685,7 +685,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask1(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[0,2],mem[3,2],zmm0[4,6],mem[7,6],zmm0[8,10],mem[11,10],zmm0[12,14],mem[15,14]
; CHECK-NEXT: retq
@ -699,7 +699,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask1(<16 x float> %vec
define <16 x float> @test_16xfloat_masked_shuff_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[2,0],mem[2,2],zmm0[6,4],mem[6,6],zmm0[10,8],mem[10,10],zmm0[14,12],mem[14,14]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -714,7 +714,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask2(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,0],mem[2,2],zmm0[6,4],mem[6,6],zmm0[10,8],mem[10,10],zmm0[14,12],mem[14,14]
; CHECK-NEXT: retq
@ -737,7 +737,7 @@ define <16 x float> @test_16xfloat_shuff_mem_mask3(<16 x float> %vec1, <16 x flo
define <16 x float> @test_16xfloat_masked_shuff_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm1 {%k1} = zmm0[2,1],mem[1,3],zmm0[6,5],mem[5,7],zmm0[10,9],mem[9,11],zmm0[14,13],mem[13,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -752,7 +752,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask3(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufps {{.*#+}} zmm0 {%k1} {z} = zmm0[2,1],mem[1,3],zmm0[6,5],mem[5,7],zmm0[10,9],mem[9,11],zmm0[14,13],mem[13,15]
; CHECK-NEXT: retq
@ -1155,7 +1155,7 @@ define <8 x double> @test_8xdouble_shuff_mask0(<8 x double> %vec1, <8 x double>
define <8 x double> @test_8xdouble_masked_shuff_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[4],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1169,7 +1169,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask0(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[4],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -1181,7 +1181,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mask0(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_masked_shuff_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[5],zmm1[5],zmm0[6],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1195,7 +1195,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask1(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[3],zmm0[5],zmm1[5],zmm0[6],zmm1[7]
; CHECK-NEXT: retq
@ -1207,7 +1207,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mask1(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_masked_shuff_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[5],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1221,7 +1221,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask2(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[5],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
@ -1241,7 +1241,7 @@ define <8 x double> @test_8xdouble_shuff_mask3(<8 x double> %vec1, <8 x double>
define <8 x double> @test_8xdouble_masked_shuff_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1255,7 +1255,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask3(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[0],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -1276,7 +1276,7 @@ define <8 x double> @test_8xdouble_shuff_mem_mask0(<8 x double> %vec1, <8 x doub
define <8 x double> @test_8xdouble_masked_shuff_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[5],mem[5],zmm0[6],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1291,7 +1291,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask0(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[5],mem[5],zmm0[6],mem[7]
; CHECK-NEXT: retq
@ -1305,7 +1305,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask0(<8 x double> %vec
define <8 x double> @test_8xdouble_masked_shuff_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[0],zmm0[3],mem[2],zmm0[4],mem[4],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1320,7 +1320,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask1(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[0],zmm0[3],mem[2],zmm0[4],mem[4],zmm0[7],mem[7]
; CHECK-NEXT: retq
@ -1334,7 +1334,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask1(<8 x double> %vec
define <8 x double> @test_8xdouble_masked_shuff_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[2],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1349,7 +1349,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask2(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[2],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
@ -1372,7 +1372,7 @@ define <8 x double> @test_8xdouble_shuff_mem_mask3(<8 x double> %vec1, <8 x doub
define <8 x double> @test_8xdouble_masked_shuff_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[2],mem[3],zmm0[4],mem[5],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1387,7 +1387,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask3(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshufpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[2],mem[3],zmm0[4],mem[5],zmm0[6],mem[6]
; CHECK-NEXT: retq

64
test/CodeGen/X86/avx512-shuffles/shuffle-vec.ll
View File

@ -276,7 +276,7 @@ define <16 x float> @test_16xfloat_shuff_mask0(<16 x float> %vec1, <16 x float>
define <16 x float> @test_16xfloat_masked_shuff_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[12,13,14,15,0,1,2,3],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -290,7 +290,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask0(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,0,1,2,3],zmm1[4,5,6,7,12,13,14,15]
; CHECK-NEXT: retq
@ -302,7 +302,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mask0(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_masked_shuff_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[0,1,2,3,8,9,10,11],zmm1[0,1,2,3,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -316,7 +316,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask1(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,8,9,10,11],zmm1[0,1,2,3,12,13,14,15]
; CHECK-NEXT: retq
@ -328,7 +328,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mask1(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_masked_shuff_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[12,13,14,15,4,5,6,7],zmm1[0,1,2,3,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -342,7 +342,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask2(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,4,5,6,7],zmm1[0,1,2,3,4,5,6,7]
; CHECK-NEXT: retq
@ -362,7 +362,7 @@ define <16 x float> @test_16xfloat_shuff_mask3(<16 x float> %vec1, <16 x float>
define <16 x float> @test_16xfloat_masked_shuff_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm2 {%k1} = zmm0[8,9,10,11,12,13,14,15],zmm1[0,1,2,3,8,9,10,11]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -376,7 +376,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mask3(<16 x float> %vec1, <16 x
define <16 x float> @test_16xfloat_zero_masked_shuff_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,12,13,14,15],zmm1[0,1,2,3,8,9,10,11]
; CHECK-NEXT: retq
@ -397,7 +397,7 @@ define <16 x float> @test_16xfloat_shuff_mem_mask0(<16 x float> %vec1, <16 x flo
define <16 x float> @test_16xfloat_masked_shuff_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[12,13,14,15,8,9,10,11],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -412,7 +412,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask0(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[12,13,14,15,8,9,10,11],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
@ -426,7 +426,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask0(<16 x float> %vec
define <16 x float> @test_16xfloat_masked_shuff_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -441,7 +441,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask1(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[8,9,10,11,4,5,6,7],mem[8,9,10,11,4,5,6,7]
; CHECK-NEXT: retq
@ -455,7 +455,7 @@ define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask1(<16 x float> %vec
define <16 x float> @test_16xfloat_masked_shuff_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3,0,1,2,3],mem[8,9,10,11,8,9,10,11]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -470,7 +470,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask2(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3,0,1,2,3],mem[8,9,10,11,8,9,10,11]
; CHECK-NEXT: retq
@ -493,7 +493,7 @@ define <16 x float> @test_16xfloat_shuff_mem_mask3(<16 x float> %vec1, <16 x flo
define <16 x float> @test_16xfloat_masked_shuff_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm1 {%k1} = zmm0[4,5,6,7,0,1,2,3],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -508,7 +508,7 @@ define <16 x float> @test_16xfloat_masked_shuff_mem_mask3(<16 x float> %vec1, <1
define <16 x float> @test_16xfloat_zero_masked_shuff_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff32x4 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,6,7,0,1,2,3],mem[12,13,14,15,12,13,14,15]
; CHECK-NEXT: retq
@ -792,7 +792,7 @@ define <8 x double> @test_8xdouble_shuff_mask0(<8 x double> %vec1, <8 x double>
define <8 x double> @test_8xdouble_masked_shuff_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[6,7,2,3],zmm1[6,7,0,1]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -806,7 +806,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask0(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,2,3],zmm1[6,7,0,1]
; CHECK-NEXT: retq
@ -818,7 +818,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mask0(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_masked_shuff_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[0,1,4,5],zmm1[0,1,4,5]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -832,7 +832,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask1(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,4,5],zmm1[0,1,4,5]
; CHECK-NEXT: retq
@ -844,7 +844,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mask1(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_masked_shuff_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[6,7,4,5],zmm1[4,5,0,1]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -858,7 +858,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask2(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,4,5],zmm1[4,5,0,1]
; CHECK-NEXT: retq
@ -878,7 +878,7 @@ define <8 x double> @test_8xdouble_shuff_mask3(<8 x double> %vec1, <8 x double>
define <8 x double> @test_8xdouble_masked_shuff_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm2 {%k1} = zmm0[4,5,4,5],zmm1[4,5,2,3]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -892,7 +892,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mask3(<8 x double> %vec1, <8 x d
define <8 x double> @test_8xdouble_zero_masked_shuff_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[4,5,4,5],zmm1[4,5,2,3]
; CHECK-NEXT: retq
@ -913,7 +913,7 @@ define <8 x double> @test_8xdouble_shuff_mem_mask0(<8 x double> %vec1, <8 x doub
define <8 x double> @test_8xdouble_masked_shuff_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[6,7,0,1],mem[0,1,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -928,7 +928,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask0(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,0,1],mem[0,1,0,1]
; CHECK-NEXT: retq
@ -942,7 +942,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask0(<8 x double> %vec
define <8 x double> @test_8xdouble_masked_shuff_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[6,7,6,7],mem[0,1,2,3]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -957,7 +957,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask1(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[6,7,6,7],mem[0,1,2,3]
; CHECK-NEXT: retq
@ -971,7 +971,7 @@ define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask1(<8 x double> %vec
define <8 x double> @test_8xdouble_masked_shuff_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[0,1,2,3],mem[0,1,4,5]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -986,7 +986,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask2(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[0,1,2,3],mem[0,1,4,5]
; CHECK-NEXT: retq
@ -1009,7 +1009,7 @@ define <8 x double> @test_8xdouble_shuff_mem_mask3(<8 x double> %vec1, <8 x doub
define <8 x double> @test_8xdouble_masked_shuff_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm1 {%k1} = zmm0[2,3,0,1],mem[4,5,0,1]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1024,7 +1024,7 @@ define <8 x double> @test_8xdouble_masked_shuff_mem_mask3(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_zero_masked_shuff_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_shuff_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vshuff64x2 {{.*#+}} zmm0 {%k1} {z} = zmm0[2,3,0,1],mem[4,5,0,1]
; CHECK-NEXT: retq

128
test/CodeGen/X86/avx512-shuffles/unpack.ll
View File

@ -520,7 +520,7 @@ define <16 x float> @test_16xfloat_unpack_low_mask0(<16 x float> %vec1, <16 x fl
define <16 x float> @test_16xfloat_masked_unpack_low_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -534,7 +534,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mask0(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
@ -546,7 +546,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask0(<16 x float> %ve
define <16 x float> @test_16xfloat_masked_unpack_low_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -560,7 +560,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mask1(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
@ -572,7 +572,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask1(<16 x float> %ve
define <16 x float> @test_16xfloat_masked_unpack_low_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -586,7 +586,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mask2(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
@ -606,7 +606,7 @@ define <16 x float> @test_16xfloat_unpack_low_mask3(<16 x float> %vec1, <16 x fl
define <16 x float> @test_16xfloat_masked_unpack_low_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -620,7 +620,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mask3(<16 x float> %vec1, <
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13]
; CHECK-NEXT: retq
@ -641,7 +641,7 @@ define <16 x float> @test_16xfloat_unpack_low_mem_mask0(<16 x float> %vec1, <16
define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -656,7 +656,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask0(<16 x float> %vec
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
@ -670,7 +670,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask0(<16 x float>
define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -685,7 +685,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask1(<16 x float> %vec
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
@ -699,7 +699,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask1(<16 x float>
define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -714,7 +714,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask2(<16 x float> %vec
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
@ -737,7 +737,7 @@ define <16 x float> @test_16xfloat_unpack_low_mem_mask3(<16 x float> %vec1, <16
define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -752,7 +752,7 @@ define <16 x float> @test_16xfloat_masked_unpack_low_mem_mask3(<16 x float> %vec
define <16 x float> @test_16xfloat_zero_masked_unpack_low_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[1],mem[1],zmm0[4],mem[4],zmm0[5],mem[5],zmm0[8],mem[8],zmm0[9],mem[9],zmm0[12],mem[12],zmm0[13],mem[13]
; CHECK-NEXT: retq
@ -1155,7 +1155,7 @@ define <8 x double> @test_8xdouble_unpack_low_mask0(<8 x double> %vec1, <8 x dou
define <8 x double> @test_8xdouble_masked_unpack_low_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1169,7 +1169,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mask0(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
@ -1181,7 +1181,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask0(<8 x double> %ve
define <8 x double> @test_8xdouble_masked_unpack_low_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1195,7 +1195,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mask1(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
@ -1207,7 +1207,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask1(<8 x double> %ve
define <8 x double> @test_8xdouble_masked_unpack_low_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1221,7 +1221,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mask2(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
@ -1241,7 +1241,7 @@ define <8 x double> @test_8xdouble_unpack_low_mask3(<8 x double> %vec1, <8 x dou
define <8 x double> @test_8xdouble_masked_unpack_low_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 {%k1} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -1255,7 +1255,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mask3(<8 x double> %vec1, <
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6]
; CHECK-NEXT: retq
@ -1276,7 +1276,7 @@ define <8 x double> @test_8xdouble_unpack_low_mem_mask0(<8 x double> %vec1, <8 x
define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1291,7 +1291,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask0(<8 x double> %vec
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
@ -1305,7 +1305,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask0(<8 x double>
define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1320,7 +1320,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask1(<8 x double> %vec
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
@ -1334,7 +1334,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask1(<8 x double>
define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1349,7 +1349,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask2(<8 x double> %vec
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
@ -1372,7 +1372,7 @@ define <8 x double> @test_8xdouble_unpack_low_mem_mask3(<8 x double> %vec1, <8 x
define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm1 {%k1} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -1387,7 +1387,7 @@ define <8 x double> @test_8xdouble_masked_unpack_low_mem_mask3(<8 x double> %vec
define <8 x double> @test_8xdouble_zero_masked_unpack_low_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_low_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 {%k1} {z} = zmm0[0],mem[0],zmm0[2],mem[2],zmm0[4],mem[4],zmm0[6],mem[6]
; CHECK-NEXT: retq
@ -1917,7 +1917,7 @@ define <16 x float> @test_16xfloat_unpack_high_mask0(<16 x float> %vec1, <16 x f
define <16 x float> @test_16xfloat_masked_unpack_high_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -1931,7 +1931,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mask0(<16 x float> %vec1,
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask0(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
@ -1943,7 +1943,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask0(<16 x float> %v
define <16 x float> @test_16xfloat_masked_unpack_high_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -1957,7 +1957,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mask1(<16 x float> %vec1,
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask1(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
@ -1969,7 +1969,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask1(<16 x float> %v
define <16 x float> @test_16xfloat_masked_unpack_high_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -1983,7 +1983,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mask2(<16 x float> %vec1,
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask2(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
@ -2003,7 +2003,7 @@ define <16 x float> @test_16xfloat_unpack_high_mask3(<16 x float> %vec1, <16 x f
define <16 x float> @test_16xfloat_masked_unpack_high_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqps %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 {%k1} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: vmovaps %zmm2, %zmm0
@ -2017,7 +2017,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mask3(<16 x float> %vec1,
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mask3(<16 x float> %vec1, <16 x float> %vec2, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15]
; CHECK-NEXT: retq
@ -2038,7 +2038,7 @@ define <16 x float> @test_16xfloat_unpack_high_mem_mask0(<16 x float> %vec1, <16
define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2053,7 +2053,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask0(<16 x float> %ve
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask0(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
@ -2067,7 +2067,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask0(<16 x float
define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2082,7 +2082,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask1(<16 x float> %ve
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask1(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
@ -2096,7 +2096,7 @@ define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask1(<16 x float
define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2111,7 +2111,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask2(<16 x float> %ve
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask2(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
@ -2134,7 +2134,7 @@ define <16 x float> @test_16xfloat_unpack_high_mem_mask3(<16 x float> %vec1, <16
define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %vec3, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqps %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm1 {%k1} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: vmovaps %zmm1, %zmm0
@ -2149,7 +2149,7 @@ define <16 x float> @test_16xfloat_masked_unpack_high_mem_mask3(<16 x float> %ve
define <16 x float> @test_16xfloat_zero_masked_unpack_high_mem_mask3(<16 x float> %vec1, <16 x float>* %vec2p, <16 x float> %mask) {
; CHECK-LABEL: test_16xfloat_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqps %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 {%k1} {z} = zmm0[2],mem[2],zmm0[3],mem[3],zmm0[6],mem[6],zmm0[7],mem[7],zmm0[10],mem[10],zmm0[11],mem[11],zmm0[14],mem[14],zmm0[15],mem[15]
; CHECK-NEXT: retq
@ -2552,7 +2552,7 @@ define <8 x double> @test_8xdouble_unpack_high_mask0(<8 x double> %vec1, <8 x do
define <8 x double> @test_8xdouble_masked_unpack_high_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -2566,7 +2566,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mask0(<8 x double> %vec1,
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask0(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -2578,7 +2578,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask0(<8 x double> %v
define <8 x double> @test_8xdouble_masked_unpack_high_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -2592,7 +2592,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mask1(<8 x double> %vec1,
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask1(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -2604,7 +2604,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask1(<8 x double> %v
define <8 x double> @test_8xdouble_masked_unpack_high_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -2618,7 +2618,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mask2(<8 x double> %vec1,
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask2(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -2638,7 +2638,7 @@ define <8 x double> @test_8xdouble_unpack_high_mask3(<8 x double> %vec1, <8 x do
define <8 x double> @test_8xdouble_masked_unpack_high_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vpxor %xmm4, %xmm4, %xmm4
; CHECK-NEXT: vcmpeqpd %zmm4, %zmm3, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 {%k1} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: vmovapd %zmm2, %zmm0
@ -2652,7 +2652,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mask3(<8 x double> %vec1,
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mask3(<8 x double> %vec1, <8 x double> %vec2, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7]
; CHECK-NEXT: retq
@ -2673,7 +2673,7 @@ define <8 x double> @test_8xdouble_unpack_high_mem_mask0(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2688,7 +2688,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask0(<8 x double> %ve
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask0(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask0:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
@ -2702,7 +2702,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask0(<8 x double
define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2717,7 +2717,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask1(<8 x double> %ve
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask1(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask1:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
@ -2731,7 +2731,7 @@ define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask1(<8 x double
define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2746,7 +2746,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask2(<8 x double> %ve
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask2(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask2:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq
@ -2769,7 +2769,7 @@ define <8 x double> @test_8xdouble_unpack_high_mem_mask3(<8 x double> %vec1, <8
define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %vec3, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vpxor %xmm3, %xmm3, %xmm3
; CHECK-NEXT: vcmpeqpd %zmm3, %zmm2, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm1 {%k1} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: vmovapd %zmm1, %zmm0
@ -2784,7 +2784,7 @@ define <8 x double> @test_8xdouble_masked_unpack_high_mem_mask3(<8 x double> %ve
define <8 x double> @test_8xdouble_zero_masked_unpack_high_mem_mask3(<8 x double> %vec1, <8 x double>* %vec2p, <8 x double> %mask) {
; CHECK-LABEL: test_8xdouble_zero_masked_unpack_high_mem_mask3:
; CHECK: # BB#0:
; CHECK-NEXT: vxorpd %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vcmpeqpd %zmm2, %zmm1, %k1
; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 {%k1} {z} = zmm0[1],mem[1],zmm0[3],mem[3],zmm0[5],mem[5],zmm0[7],mem[7]
; CHECK-NEXT: retq