; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx512bw | FileCheck %s declare <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8>, <64 x i8>, <64 x i8>, i64) declare <16 x float> @llvm.x86.avx512.mask.vpermilvar.ps.512(<16 x float>, <16 x i32>, <16 x float>, i16) declare <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double>, <8 x i64>, <8 x double>, i8) declare <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) declare <16 x i32> @llvm.x86.avx512.mask.permvar.si.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) declare <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) declare <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64>, <8 x double>, <8 x double>, i8) declare <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32>, <16 x float>, <16 x float>, i16) declare <8 x i64> @llvm.x86.avx512.maskz.vpermt2var.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) declare <16 x i32> @llvm.x86.avx512.maskz.vpermt2var.d.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) declare <32 x i16> @llvm.x86.avx512.maskz.vpermt2var.hi.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) define <8 x double> @combine_permvar_8f64_identity(<8 x double> %x0, <8 x double> %x1) { ; CHECK-LABEL: combine_permvar_8f64_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %x0, <8 x i64> , <8 x double> %x1, i8 -1) %res1 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %res0, <8 x i64> , <8 x double> %res0, i8 -1) ret <8 x double> %res1 } define <8 x double> @combine_permvar_8f64_identity_mask(<8 x double> %x0, <8 x double> %x1, i8 %m) { ; CHECK-LABEL: combine_permvar_8f64_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm2 = [7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermpd %zmm0, %zmm2, %zmm1 {%k1} ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm0 = [7,14,5,12,3,10,1,8] ; CHECK-NEXT: vpermpd %zmm1, %zmm0, %zmm1 {%k1} ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %x0, <8 x i64> , <8 x double> %x1, i8 %m) %res1 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %res0, <8 x i64> , <8 x double> %res0, i8 %m) ret <8 x double> %res1 } define <8 x i64> @combine_permvar_8i64_identity(<8 x i64> %x0, <8 x i64> %x1) { ; CHECK-LABEL: combine_permvar_8i64_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %x0, <8 x i64> , <8 x i64> %x1, i8 -1) %res1 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %res0, <8 x i64> , <8 x i64> %res0, i8 -1) ret <8 x i64> %res1 } define <8 x i64> @combine_permvar_8i64_identity_mask(<8 x i64> %x0, <8 x i64> %x1, i8 %m) { ; CHECK-LABEL: combine_permvar_8i64_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm2 = [7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermq %zmm0, %zmm2, %zmm1 {%k1} ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm0 = [7,14,5,12,3,10,1,8] ; CHECK-NEXT: vpermq %zmm1, %zmm0, %zmm1 {%k1} ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %res0 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %x0, <8 x i64> , <8 x i64> %x1, i8 %m) %res1 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %res0, <8 x i64> , <8 x i64> %res0, i8 %m) ret <8 x i64> %res1 } define <8 x double> @combine_vpermt2var_8f64_identity(<8 x double> %x0, <8 x double> %x1) { ; CHECK-LABEL: combine_vpermt2var_8f64_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %x0, <8 x double> %x1, i8 -1) %res1 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %res0, <8 x double> %res0, i8 -1) ret <8 x double> %res1 } define <8 x double> @combine_vpermt2var_8f64_identity_mask(<8 x double> %x0, <8 x double> %x1, i8 %m) { ; CHECK-LABEL: combine_vpermt2var_8f64_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm2 = [7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermt2pd %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm1 = [7,14,5,12,3,10,1,8] ; CHECK-NEXT: vpermt2pd %zmm0, %zmm1, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %x0, <8 x double> %x1, i8 %m) %res1 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %res0, <8 x double> %res0, i8 %m) ret <8 x double> %res1 } define <8 x double> @combine_vpermt2var_8f64_movddup(<8 x double> %x0, <8 x double> %x1) { ; CHECK-LABEL: combine_vpermt2var_8f64_movddup: ; CHECK: # BB#0: ; CHECK-NEXT: vmovddup {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6] ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %x0, <8 x double> %x1, i8 -1) ret <8 x double> %res0 } define <8 x double> @combine_vpermt2var_8f64_movddup_load(<8 x double> *%p0, <8 x double> %x1) { ; CHECK-LABEL: combine_vpermt2var_8f64_movddup_load: ; CHECK: # BB#0: ; CHECK-NEXT: vmovddup {{.*#+}} zmm0 = mem[0,0,2,2,4,4,6,6] ; CHECK-NEXT: retq %x0 = load <8 x double>, <8 x double> *%p0 %res0 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %x0, <8 x double> %x1, i8 -1) ret <8 x double> %res0 } define <8 x double> @combine_vpermt2var_8f64_movddup_mask(<8 x double> %x0, <8 x double> %x1, i8 %m) { ; CHECK-LABEL: combine_vpermt2var_8f64_movddup_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovddup {{.*#+}} zmm0 {%k1} {z} = zmm0[0,0,2,2,4,4,6,6] ; CHECK-NEXT: retq %res0 = call <8 x double> @llvm.x86.avx512.maskz.vpermt2var.pd.512(<8 x i64> , <8 x double> %x0, <8 x double> %x1, i8 %m) ret <8 x double> %res0 } define <8 x i64> @combine_vpermt2var_8i64_identity(<8 x i64> %x0, <8 x i64> %x1) { ; CHECK-LABEL: combine_vpermt2var_8i64_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <8 x i64> @llvm.x86.avx512.maskz.vpermt2var.q.512(<8 x i64> , <8 x i64> %x0, <8 x i64> %x1, i8 -1) %res1 = call <8 x i64> @llvm.x86.avx512.maskz.vpermt2var.q.512(<8 x i64> , <8 x i64> %res0, <8 x i64> %res0, i8 -1) ret <8 x i64> %res1 } define <8 x i64> @combine_vpermt2var_8i64_identity_mask(<8 x i64> %x0, <8 x i64> %x1, i8 %m) { ; CHECK-LABEL: combine_vpermt2var_8i64_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm2 = [7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermt2q %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: vmovdqa64 {{.*#+}} zmm1 = [7,14,5,12,3,10,1,8] ; CHECK-NEXT: vpermt2q %zmm0, %zmm1, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <8 x i64> @llvm.x86.avx512.maskz.vpermt2var.q.512(<8 x i64> , <8 x i64> %x0, <8 x i64> %x1, i8 %m) %res1 = call <8 x i64> @llvm.x86.avx512.maskz.vpermt2var.q.512(<8 x i64> , <8 x i64> %res0, <8 x i64> %res0, i8 %m) ret <8 x i64> %res1 } define <16 x float> @combine_vpermt2var_16f32_identity(<16 x float> %x0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) %res1 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %res0, <16 x float> %res0, i16 -1) ret <16 x float> %res1 } define <16 x float> @combine_vpermt2var_16f32_identity_mask(<16 x float> %x0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermt2ps %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [15,30,13,28,11,26,9,24,7,22,5,20,3,18,1,16] ; CHECK-NEXT: vpermt2ps %zmm0, %zmm1, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) %res1 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %res0, <16 x float> %res0, i16 %m) ret <16 x float> %res1 } define <16 x float> @combine_vpermt2var_16f32_vmovddup(<16 x float> %x0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovddup: ; CHECK: # BB#0: ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,1,0,1,4,5,4,5,8,9,8,9,12,13,12,13] ; CHECK-NEXT: vpermt2ps %zmm1, %zmm2, %zmm0 ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovddup_load(<16 x float> *%p0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovddup_load: ; CHECK: # BB#0: ; CHECK-NEXT: vmovaps (%rdi), %zmm1 ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,1,0,1,4,5,4,5,8,9,8,9,12,13,12,13] ; CHECK-NEXT: vpermt2ps %zmm0, %zmm2, %zmm1 ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovddup_mask(<16 x float> %x0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovddup_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,1,0,1,4,5,4,5,8,9,8,9,12,13,12,13] ; CHECK-NEXT: vpermt2ps %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovddup_mask_load(<16 x float> *%p0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovddup_mask_load: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %esi, %k1 ; CHECK-NEXT: vmovaps (%rdi), %zmm1 ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [0,1,0,1,4,5,4,5,8,9,8,9,12,13,12,13] ; CHECK-NEXT: vpermt2ps %zmm0, %zmm2, %zmm1 {%k1} {z} ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovshdup(<16 x float> %x0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovshdup: ; CHECK: # BB#0: ; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 = zmm0[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15] ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovshdup_load(<16 x float> *%p0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovshdup_load: ; CHECK: # BB#0: ; CHECK-NEXT: vmovshdup {{.*#+}} zmm0 = mem[1,1,3,3,5,5,7,7,9,9,11,11,13,13,15,15] ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovshdup_mask(<16 x float> %x0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovshdup_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %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 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovsldup(<16 x float> %x0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovsldup: ; CHECK: # BB#0: ; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 = zmm0[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14] ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovsldup_load(<16 x float> *%p0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovsldup_load: ; CHECK: # BB#0: ; CHECK-NEXT: vmovsldup {{.*#+}} zmm0 = mem[0,0,2,2,4,4,6,6,8,8,10,10,12,12,14,14] ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovsldup_mask(<16 x float> %x0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovsldup_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %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 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vmovsldup_mask_load(<16 x float> *%p0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vmovsldup_mask_load: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %esi, %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 %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vpermilps(<16 x float> %x0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vpermilps: ; CHECK: # BB#0: ; CHECK-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12] ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vpermilps_load(<16 x float> *%p0, <16 x float> %x1) { ; CHECK-LABEL: combine_vpermt2var_16f32_vpermilps_load: ; CHECK: # BB#0: ; CHECK-NEXT: vpermilps {{.*#+}} zmm0 = mem[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12] ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 -1) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vpermilps_mask(<16 x float> %x0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vpermilps_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = zmm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12] ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x float> @combine_vpermt2var_16f32_vpermilps_mask_load(<16 x float> *%p0, <16 x float> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16f32_vpermilps_mask_load: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %esi, %k1 ; CHECK-NEXT: vpermilps {{.*#+}} zmm0 {%k1} {z} = mem[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12] ; CHECK-NEXT: retq %x0 = load <16 x float>, <16 x float> *%p0 %res0 = call <16 x float> @llvm.x86.avx512.maskz.vpermt2var.ps.512(<16 x i32> , <16 x float> %x0, <16 x float> %x1, i16 %m) ret <16 x float> %res0 } define <16 x i32> @combine_vpermt2var_16i32_identity(<16 x i32> %x0, <16 x i32> %x1) { ; CHECK-LABEL: combine_vpermt2var_16i32_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <16 x i32> @llvm.x86.avx512.maskz.vpermt2var.d.512(<16 x i32> , <16 x i32> %x0, <16 x i32> %x1, i16 -1) %res1 = call <16 x i32> @llvm.x86.avx512.maskz.vpermt2var.d.512(<16 x i32> , <16 x i32> %res0, <16 x i32> %res0, i16 -1) ret <16 x i32> %res1 } define <16 x i32> @combine_vpermt2var_16i32_identity_mask(<16 x i32> %x0, <16 x i32> %x1, i16 %m) { ; CHECK-LABEL: combine_vpermt2var_16i32_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm2 = [15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermt2d %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: vmovdqa32 {{.*#+}} zmm1 = [15,30,13,28,11,26,9,24,7,22,5,20,3,18,1,16] ; CHECK-NEXT: vpermt2d %zmm0, %zmm1, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <16 x i32> @llvm.x86.avx512.maskz.vpermt2var.d.512(<16 x i32> , <16 x i32> %x0, <16 x i32> %x1, i16 %m) %res1 = call <16 x i32> @llvm.x86.avx512.maskz.vpermt2var.d.512(<16 x i32> , <16 x i32> %res0, <16 x i32> %res0, i16 %m) ret <16 x i32> %res1 } define <32 x i16> @combine_vpermt2var_32i16_identity(<32 x i16> %x0, <32 x i16> %x1) { ; CHECK-LABEL: combine_vpermt2var_32i16_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %res0 = call <32 x i16> @llvm.x86.avx512.maskz.vpermt2var.hi.512(<32 x i16> , <32 x i16> %x0, <32 x i16> %x1, i32 -1) %res1 = call <32 x i16> @llvm.x86.avx512.maskz.vpermt2var.hi.512(<32 x i16> , <32 x i16> %res0, <32 x i16> %res0, i32 -1) ret <32 x i16> %res1 } define <32 x i16> @combine_vpermt2var_32i16_identity_mask(<32 x i16> %x0, <32 x i16> %x1, i32 %m) { ; CHECK-LABEL: combine_vpermt2var_32i16_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovd %edi, %k1 ; CHECK-NEXT: vmovdqu16 {{.*#+}} zmm2 = [31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpermt2w %zmm1, %zmm2, %zmm0 {%k1} {z} ; CHECK-NEXT: vmovdqu16 {{.*#+}} zmm1 = [63,30,61,28,59,26,57,24,55,22,53,20,51,18,49,16,47,46,13,44,11,42,9,40,7,38,5,36,3,34,1,32] ; CHECK-NEXT: vpermt2w %zmm0, %zmm1, %zmm0 {%k1} {z} ; CHECK-NEXT: retq %res0 = call <32 x i16> @llvm.x86.avx512.maskz.vpermt2var.hi.512(<32 x i16> , <32 x i16> %x0, <32 x i16> %x1, i32 %m) %res1 = call <32 x i16> @llvm.x86.avx512.maskz.vpermt2var.hi.512(<32 x i16> , <32 x i16> %res0, <32 x i16> %res0, i32 %m) ret <32 x i16> %res1 } define <64 x i8> @combine_pshufb_identity(<64 x i8> %x0) { ; CHECK-LABEL: combine_pshufb_identity: ; CHECK: # BB#0: ; CHECK-NEXT: retq %select = bitcast <8 x i64> to <64 x i8> %mask = bitcast <16 x i32> to <64 x i8> %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %x0, <64 x i8> %mask, <64 x i8> %select, i64 -1) %res1 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %res0, <64 x i8> %mask, <64 x i8> %select, i64 -1) ret <64 x i8> %res1 } define <64 x i8> @combine_pshufb_identity_mask(<64 x i8> %x0, i64 %m) { ; CHECK-LABEL: combine_pshufb_identity_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovq %rdi, %k1 ; CHECK-NEXT: vpternlogd $255, %zmm1, %zmm1, %zmm1 ; CHECK-NEXT: vmovdqu8 {{.*#+}} zmm2 = [15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] ; CHECK-NEXT: vpternlogd $255, %zmm3, %zmm3, %zmm3 ; CHECK-NEXT: vpshufb %zmm2, %zmm0, %zmm3 {%k1} ; CHECK-NEXT: vpshufb %zmm2, %zmm3, %zmm1 {%k1} ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %select = bitcast <8 x i64> to <64 x i8> %mask = bitcast <16 x i32> to <64 x i8> %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %x0, <64 x i8> %mask, <64 x i8> %select, i64 %m) %res1 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %res0, <64 x i8> %mask, <64 x i8> %select, i64 %m) ret <64 x i8> %res1 } define <32 x i16> @combine_permvar_as_vpbroadcastw512(<32 x i16> %x0) { ; CHECK-LABEL: combine_permvar_as_vpbroadcastw512: ; CHECK: # BB#0: ; CHECK-NEXT: vpbroadcastw %xmm0, %zmm0 ; CHECK-NEXT: retq %1 = call <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16> %x0, <32 x i16> zeroinitializer, <32 x i16> undef, i32 -1) ret <32 x i16> %1 } define <16 x i32> @combine_permvar_as_vpbroadcastd512(<16 x i32> %x0) { ; CHECK-LABEL: combine_permvar_as_vpbroadcastd512: ; CHECK: # BB#0: ; CHECK-NEXT: vpbroadcastd %xmm0, %zmm0 ; CHECK-NEXT: retq %1 = call <16 x i32> @llvm.x86.avx512.mask.permvar.si.512(<16 x i32> %x0, <16 x i32> zeroinitializer, <16 x i32> undef, i16 -1) ret <16 x i32> %1 } define <8 x i64> @combine_permvar_as_vpbroadcastq512(<8 x i64> %x0) { ; CHECK-LABEL: combine_permvar_as_vpbroadcastq512: ; CHECK: # BB#0: ; CHECK-NEXT: vpbroadcastq %xmm0, %zmm0 ; CHECK-NEXT: retq %1 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %x0, <8 x i64> zeroinitializer, <8 x i64> undef, i8 -1) ret <8 x i64> %1 } define <8 x i64> @combine_permvar_8i64_as_permq(<8 x i64> %x0, <8 x i64> %x1) { ; CHECK-LABEL: combine_permvar_8i64_as_permq: ; CHECK: # BB#0: ; CHECK-NEXT: vpermq {{.*#+}} zmm0 = zmm0[3,2,1,0,7,6,5,4] ; CHECK-NEXT: retq %1 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %x0, <8 x i64> , <8 x i64> %x1, i8 -1) ret <8 x i64> %1 } define <8 x i64> @combine_permvar_8i64_as_permq_mask(<8 x i64> %x0, <8 x i64> %x1, i8 %m) { ; CHECK-LABEL: combine_permvar_8i64_as_permq_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vpermq {{.*#+}} zmm1 {%k1} = zmm0[3,2,1,0,7,6,5,4] ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %1 = call <8 x i64> @llvm.x86.avx512.mask.permvar.di.512(<8 x i64> %x0, <8 x i64> , <8 x i64> %x1, i8 %m) ret <8 x i64> %1 } define <8 x double> @combine_permvar_8f64_as_permpd(<8 x double> %x0, <8 x double> %x1) { ; CHECK-LABEL: combine_permvar_8f64_as_permpd: ; CHECK: # BB#0: ; CHECK-NEXT: vpermpd {{.*#+}} zmm0 = zmm0[3,2,1,0,7,6,5,4] ; CHECK-NEXT: retq %1 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %x0, <8 x i64> , <8 x double> %x1, i8 -1) ret <8 x double> %1 } define <8 x double> @combine_permvar_8f64_as_permpd_mask(<8 x double> %x0, <8 x double> %x1, i8 %m) { ; CHECK-LABEL: combine_permvar_8f64_as_permpd_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovw %edi, %k1 ; CHECK-NEXT: vpermpd {{.*#+}} zmm1 {%k1} = zmm0[3,2,1,0,7,6,5,4] ; CHECK-NEXT: vmovaps %zmm1, %zmm0 ; CHECK-NEXT: retq %1 = call <8 x double> @llvm.x86.avx512.mask.permvar.df.512(<8 x double> %x0, <8 x i64> , <8 x double> %x1, i8 %m) ret <8 x double> %1 } define <16 x float> @combine_vpermilvar_16f32_230146759A8BCFDE(<16 x float> %x0) { ; CHECK-LABEL: combine_vpermilvar_16f32_230146759A8BCFDE: ; CHECK: # BB#0: ; CHECK-NEXT: vpermilps {{.*#+}} zmm0 = zmm0[2,3,0,1,4,6,7,5,9,10,8,11,12,15,13,14] ; CHECK-NEXT: retq %res0 = call <16 x float> @llvm.x86.avx512.mask.vpermilvar.ps.512(<16 x float> %x0, <16 x i32> , <16 x float> undef, i16 -1) %res1 = call <16 x float> @llvm.x86.avx512.mask.vpermilvar.ps.512(<16 x float> %res0, <16 x i32> , <16 x float> undef, i16 -1) ret <16 x float> %res1 } define <64 x i8> @combine_pshufb_as_pslldq(<64 x i8> %a0) { ; CHECK-LABEL: combine_pshufb_as_pslldq: ; CHECK: # BB#0: ; CHECK-NEXT: vpshufb {{.*#+}} zmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[16,17,18,19,20,21],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[32,33,34,35,36,37],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[48,49,50,51,52,53] ; CHECK-NEXT: retq %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %a0, <64 x i8> , <64 x i8> undef, i64 -1) ret <64 x i8> %res0 } define <64 x i8> @combine_pshufb_as_pslldq_mask(<64 x i8> %a0, i64 %m) { ; CHECK-LABEL: combine_pshufb_as_pslldq_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovq %rdi, %k1 ; CHECK-NEXT: vpshufb {{.*#+}} zmm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[16,17,18,19,20,21],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[32,33,34,35,36,37],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[48,49,50,51,52,53] ; CHECK-NEXT: retq %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %a0, <64 x i8> , <64 x i8> zeroinitializer, i64 %m) ret <64 x i8> %res0 } define <64 x i8> @combine_pshufb_as_psrldq(<64 x i8> %a0) { ; CHECK-LABEL: combine_pshufb_as_psrldq: ; CHECK: # BB#0: ; CHECK-NEXT: vpshufb {{.*#+}} zmm0 = zmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[47],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[63],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero ; CHECK-NEXT: retq %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %a0, <64 x i8> , <64 x i8> undef, i64 -1) ret <64 x i8> %res0 } define <64 x i8> @combine_pshufb_as_psrldq_mask(<64 x i8> %a0, i64 %m) { ; CHECK-LABEL: combine_pshufb_as_psrldq_mask: ; CHECK: # BB#0: ; CHECK-NEXT: kmovq %rdi, %k1 ; CHECK-NEXT: vpshufb {{.*#+}} zmm0 = zmm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[47],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zmm0[63],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero ; CHECK-NEXT: retq %res0 = call <64 x i8> @llvm.x86.avx512.mask.pshuf.b.512(<64 x i8> %a0, <64 x i8> , <64 x i8> zeroinitializer, i64 %m) ret <64 x i8> %res0 } define <32 x i16> @combine_permvar_as_pshuflw(<32 x i16> %a0) { ; CHECK-LABEL: combine_permvar_as_pshuflw: ; CHECK: # BB#0: ; CHECK-NEXT: vpshuflw {{.*#+}} zmm0 = zmm0[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15,17,16,19,18,20,21,22,23,25,24,27,26,28,29,30,31] ; CHECK-NEXT: retq %res0 = call <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16> %a0, <32 x i16> , <32 x i16> undef, i32 -1) ret <32 x i16> %res0 } define <32 x i16> @combine_pshufb_as_pshufhw(<32 x i16> %a0) { ; CHECK-LABEL: combine_pshufb_as_pshufhw: ; CHECK: # BB#0: ; CHECK-NEXT: vpshufhw {{.*#+}} zmm0 = zmm0[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14,16,17,18,19,21,20,23,22,24,25,26,27,29,28,31,30] ; CHECK-NEXT: retq %res0 = call <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16> %a0, <32 x i16> , <32 x i16> undef, i32 -1) ret <32 x i16> %res0 } define <32 x i16> @combine_pshufb_as_pshufw(<32 x i16> %a0) { ; CHECK-LABEL: combine_pshufb_as_pshufw: ; CHECK: # BB#0: ; CHECK-NEXT: vpshuflw {{.*#+}} zmm0 = zmm0[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15,17,16,19,18,20,21,22,23,25,24,27,26,28,29,30,31] ; CHECK-NEXT: vpshufhw {{.*#+}} zmm0 = zmm0[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14,16,17,18,19,21,20,23,22,24,25,26,27,29,28,31,30] ; CHECK-NEXT: retq %res0 = call <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16> %a0, <32 x i16> , <32 x i16> undef, i32 -1) %res1 = call <32 x i16> @llvm.x86.avx512.mask.permvar.hi.512(<32 x i16> %res0, <32 x i16> , <32 x i16> undef, i32 -1) ret <32 x i16> %res1 }