; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE,SSE41
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=AVX512F
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vl | FileCheck %s --check-prefixes=AVX512VL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=AVX512BW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vbmi,+avx512vbmi2 | FileCheck %s --check-prefixes=AVX512VBMI2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512VLBW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vbmi,+avx512vbmi2,+avx512vl | FileCheck %s --check-prefixes=AVX512VLVBMI2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.1-256 | FileCheck %s --check-prefixes=AVX512VLVBMI2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.1-512 | FileCheck %s --check-prefixes=AVX512VLVBMI2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefixes=XOP,XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefixes=XOP,XOPAVX2
; Just one 32-bit run to make sure we do reasonable things for i64 cases.
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=X86-SSE2
declare <2 x i64> @llvm.fshl.v2i64(<2 x i64>, <2 x i64>, <2 x i64>)
declare <4 x i32> @llvm.fshl.v4i32(<4 x i32>, <4 x i32>, <4 x i32>)
declare <8 x i16> @llvm.fshl.v8i16(<8 x i16>, <8 x i16>, <8 x i16>)
declare <16 x i8> @llvm.fshl.v16i8(<16 x i8>, <16 x i8>, <16 x i8>)
;
; Variable Shifts
;
define <2 x i64> @var_funnnel_v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> %amt) nounwind {
; SSE2-LABEL: var_funnnel_v2i64:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [63,63]
; SSE2-NEXT: movdqa %xmm2, %xmm4
; SSE2-NEXT: pandn %xmm3, %xmm4
; SSE2-NEXT: psrlq $1, %xmm1
; SSE2-NEXT: movdqa %xmm1, %xmm5
; SSE2-NEXT: psrlq %xmm4, %xmm5
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
; SSE2-NEXT: psrlq %xmm4, %xmm1
; SSE2-NEXT: shufpd {{.*#+}} xmm5 = xmm5[0],xmm1[1]
; SSE2-NEXT: pand %xmm3, %xmm2
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: psllq %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; SSE2-NEXT: psllq %xmm2, %xmm0
; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE2-NEXT: orpd %xmm5, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_funnnel_v2i64:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovsxbq {{.*#+}} xmm3 = [63,63]
; SSE41-NEXT: movdqa %xmm2, %xmm4
; SSE41-NEXT: pandn %xmm3, %xmm4
; SSE41-NEXT: psrlq $1, %xmm1
; SSE41-NEXT: movdqa %xmm1, %xmm5
; SSE41-NEXT: psrlq %xmm4, %xmm5
; SSE41-NEXT: pshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
; SSE41-NEXT: psrlq %xmm4, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm5 = xmm5[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT: pand %xmm3, %xmm2
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: psllq %xmm2, %xmm1
; SSE41-NEXT: pshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; SSE41-NEXT: psllq %xmm2, %xmm0
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
; SSE41-NEXT: por %xmm5, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_funnnel_v2i64:
; AVX1: # %bb.0:
; AVX1-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX1-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX1-NEXT: vpsrlq %xmm4, %xmm1, %xmm5
; AVX1-NEXT: vpshufd {{.*#+}} xmm4 = xmm4[2,3,2,3]
; AVX1-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm5[0,1,2,3],xmm1[4,5,6,7]
; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX1-NEXT: vpsllq %xmm2, %xmm0, %xmm3
; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; AVX1-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm3[0,1,2,3],xmm0[4,5,6,7]
; AVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_funnnel_v2i64:
; AVX2: # %bb.0:
; AVX2-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX2-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX2-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX2-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; AVX2-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX2-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; AVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; AVX512F-LABEL: var_funnnel_v2i64:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX512F-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512F-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512F-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; AVX512F-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512F-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: var_funnnel_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpbroadcastq {{.*#+}} xmm3 = [63,63]
; AVX512VL-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VL-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; AVX512VL-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: var_funnnel_v2i64:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX512BW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512BW-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; AVX512BW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512BW-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: var_funnnel_v2i64:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm2 killed $xmm2 def $zmm2
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldvq %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: var_funnnel_v2i64:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpbroadcastq {{.*#+}} xmm3 = [63,63]
; AVX512VLBW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VLBW-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: var_funnnel_v2i64:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvq %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: var_funnnel_v2i64:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm4
; XOPAVX1-NEXT: vpshlq %xmm4, %xmm0, %xmm0
; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm2
; XOPAVX1-NEXT: vpxor %xmm3, %xmm3, %xmm3
; XOPAVX1-NEXT: vpsubq %xmm2, %xmm3, %xmm2
; XOPAVX1-NEXT: vpsrlq $1, %xmm1, %xmm1
; XOPAVX1-NEXT: vpshlq %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_funnnel_v2i64:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; XOPAVX2-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOPAVX2-NEXT: vpsrlq $1, %xmm1, %xmm1
; XOPAVX2-NEXT: vpsrlvq %xmm4, %xmm1, %xmm1
; XOPAVX2-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOPAVX2-NEXT: vpsllvq %xmm2, %xmm0, %xmm0
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: var_funnnel_v2i64:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [63,0,63,0]
; X86-SSE2-NEXT: movdqa %xmm2, %xmm5
; X86-SSE2-NEXT: pandn %xmm4, %xmm5
; X86-SSE2-NEXT: psrlq $1, %xmm1
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: psrlq %xmm5, %xmm3
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; X86-SSE2-NEXT: psrlq %xmm5, %xmm1
; X86-SSE2-NEXT: shufpd {{.*#+}} xmm3 = xmm3[0],xmm1[1]
; X86-SSE2-NEXT: pand %xmm4, %xmm2
; X86-SSE2-NEXT: movdqa %xmm0, %xmm1
; X86-SSE2-NEXT: psllq %xmm2, %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
; X86-SSE2-NEXT: psllq %xmm2, %xmm0
; X86-SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X86-SSE2-NEXT: orpd %xmm3, %xmm0
; X86-SSE2-NEXT: retl
%res = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> %amt)
ret <2 x i64> %res
}
define <4 x i32> @var_funnnel_v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> %amt) nounwind {
; SSE2-LABEL: var_funnnel_v4i32:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31]
; SSE2-NEXT: movdqa %xmm2, %xmm5
; SSE2-NEXT: pandn %xmm4, %xmm5
; SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm5[2,3,3,3,4,5,6,7]
; SSE2-NEXT: psrld $1, %xmm1
; SSE2-NEXT: movdqa %xmm1, %xmm6
; SSE2-NEXT: psrld %xmm3, %xmm6
; SSE2-NEXT: pshuflw {{.*#+}} xmm7 = xmm5[0,1,1,1,4,5,6,7]
; SSE2-NEXT: movdqa %xmm1, %xmm3
; SSE2-NEXT: psrld %xmm7, %xmm3
; SSE2-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm6[0]
; SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm5[2,3,3,3,4,5,6,7]
; SSE2-NEXT: movdqa %xmm1, %xmm7
; SSE2-NEXT: psrld %xmm6, %xmm7
; SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[0,1,1,1,4,5,6,7]
; SSE2-NEXT: psrld %xmm5, %xmm1
; SSE2-NEXT: punpckhqdq {{.*#+}} xmm1 = xmm1[1],xmm7[1]
; SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,3],xmm1[0,3]
; SSE2-NEXT: pand %xmm4, %xmm2
; SSE2-NEXT: pslld $23, %xmm2
; SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE2-NEXT: cvttps2dq %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm1, %xmm0
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE2-NEXT: por %xmm3, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_funnnel_v4i32:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovsxbd {{.*#+}} xmm3 = [31,31,31,31]
; SSE41-NEXT: movdqa %xmm2, %xmm4
; SSE41-NEXT: pandn %xmm3, %xmm4
; SSE41-NEXT: pshuflw {{.*#+}} xmm5 = xmm4[2,3,3,3,4,5,6,7]
; SSE41-NEXT: psrld $1, %xmm1
; SSE41-NEXT: movdqa %xmm1, %xmm6
; SSE41-NEXT: psrld %xmm5, %xmm6
; SSE41-NEXT: pshufd {{.*#+}} xmm5 = xmm4[2,3,2,3]
; SSE41-NEXT: pshuflw {{.*#+}} xmm7 = xmm5[2,3,3,3,4,5,6,7]
; SSE41-NEXT: movdqa %xmm1, %xmm8
; SSE41-NEXT: psrld %xmm7, %xmm8
; SSE41-NEXT: pblendw {{.*#+}} xmm8 = xmm6[0,1,2,3],xmm8[4,5,6,7]
; SSE41-NEXT: pshuflw {{.*#+}} xmm4 = xmm4[0,1,1,1,4,5,6,7]
; SSE41-NEXT: movdqa %xmm1, %xmm6
; SSE41-NEXT: psrld %xmm4, %xmm6
; SSE41-NEXT: pshuflw {{.*#+}} xmm4 = xmm5[0,1,1,1,4,5,6,7]
; SSE41-NEXT: psrld %xmm4, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm6 = xmm6[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT: pblendw {{.*#+}} xmm6 = xmm6[0,1],xmm8[2,3],xmm6[4,5],xmm8[6,7]
; SSE41-NEXT: pand %xmm3, %xmm2
; SSE41-NEXT: pslld $23, %xmm2
; SSE41-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE41-NEXT: cvttps2dq %xmm2, %xmm1
; SSE41-NEXT: pmulld %xmm1, %xmm0
; SSE41-NEXT: por %xmm6, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_funnnel_v4i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31]
; AVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX1-NEXT: vpsrldq {{.*#+}} xmm5 = xmm4[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; AVX1-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX1-NEXT: vpsrld %xmm5, %xmm1, %xmm5
; AVX1-NEXT: vpsrlq $32, %xmm4, %xmm6
; AVX1-NEXT: vpsrld %xmm6, %xmm1, %xmm6
; AVX1-NEXT: vpblendw {{.*#+}} xmm5 = xmm6[0,1,2,3],xmm5[4,5,6,7]
; AVX1-NEXT: vpxor %xmm6, %xmm6, %xmm6
; AVX1-NEXT: vpunpckhdq {{.*#+}} xmm6 = xmm4[2],xmm6[2],xmm4[3],xmm6[3]
; AVX1-NEXT: vpsrld %xmm6, %xmm1, %xmm6
; AVX1-NEXT: vpmovzxdq {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero
; AVX1-NEXT: vpsrld %xmm4, %xmm1, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm6[4,5,6,7]
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm5[2,3],xmm1[4,5],xmm5[6,7]
; AVX1-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX1-NEXT: vpslld $23, %xmm2, %xmm2
; AVX1-NEXT: vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX1-NEXT: vcvttps2dq %xmm2, %xmm2
; AVX1-NEXT: vpmulld %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_funnnel_v4i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; AVX2-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX2-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX2-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; AVX2-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX2-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; AVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; AVX512F-LABEL: var_funnnel_v4i32:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; AVX512F-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512F-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX512F-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; AVX512F-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512F-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: var_funnnel_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; AVX512VL-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VL-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; AVX512VL-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: var_funnnel_v4i32:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; AVX512BW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512BW-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; AVX512BW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512BW-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: var_funnnel_v4i32:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm2 killed $xmm2 def $zmm2
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldvd %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: var_funnnel_v4i32:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; AVX512VLBW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VLBW-NEXT: vpsrld $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: var_funnnel_v4i32:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvd %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: var_funnnel_v4i32:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [31,31,31,31]
; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm4
; XOPAVX1-NEXT: vpshld %xmm4, %xmm0, %xmm0
; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm2
; XOPAVX1-NEXT: vpxor %xmm3, %xmm3, %xmm3
; XOPAVX1-NEXT: vpsubd %xmm2, %xmm3, %xmm2
; XOPAVX1-NEXT: vpsrld $1, %xmm1, %xmm1
; XOPAVX1-NEXT: vpshld %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_funnnel_v4i32:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpbroadcastd {{.*#+}} xmm3 = [31,31,31,31]
; XOPAVX2-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOPAVX2-NEXT: vpsrld $1, %xmm1, %xmm1
; XOPAVX2-NEXT: vpsrlvd %xmm4, %xmm1, %xmm1
; XOPAVX2-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOPAVX2-NEXT: vpsllvd %xmm2, %xmm0, %xmm0
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: var_funnnel_v4i32:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [31,31,31,31]
; X86-SSE2-NEXT: movdqa %xmm2, %xmm5
; X86-SSE2-NEXT: pandn %xmm4, %xmm5
; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm5[2,3,3,3,4,5,6,7]
; X86-SSE2-NEXT: psrld $1, %xmm1
; X86-SSE2-NEXT: movdqa %xmm1, %xmm6
; X86-SSE2-NEXT: psrld %xmm3, %xmm6
; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm7 = xmm5[0,1,1,1,4,5,6,7]
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: psrld %xmm7, %xmm3
; X86-SSE2-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm6[0]
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[2,3,2,3]
; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm6 = xmm5[2,3,3,3,4,5,6,7]
; X86-SSE2-NEXT: movdqa %xmm1, %xmm7
; X86-SSE2-NEXT: psrld %xmm6, %xmm7
; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm5 = xmm5[0,1,1,1,4,5,6,7]
; X86-SSE2-NEXT: psrld %xmm5, %xmm1
; X86-SSE2-NEXT: punpckhqdq {{.*#+}} xmm1 = xmm1[1],xmm7[1]
; X86-SSE2-NEXT: shufps {{.*#+}} xmm3 = xmm3[0,3],xmm1[0,3]
; X86-SSE2-NEXT: pand %xmm4, %xmm2
; X86-SSE2-NEXT: pslld $23, %xmm2
; X86-SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: cvttps2dq %xmm2, %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm1, %xmm0
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm2, %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE2-NEXT: por %xmm3, %xmm0
; X86-SSE2-NEXT: retl
%res = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> %amt)
ret <4 x i32> %res
}
define <8 x i16> @var_funnnel_v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> %amt) nounwind {
; SSE2-LABEL: var_funnnel_v8i16:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm1, %xmm3
; SSE2-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE2-NEXT: movdqa %xmm2, %xmm4
; SSE2-NEXT: punpckhwd {{.*#+}} xmm4 = xmm4[4,4,5,5,6,6,7,7]
; SSE2-NEXT: pslld $23, %xmm4
; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [1065353216,1065353216,1065353216,1065353216]
; SSE2-NEXT: paddd %xmm5, %xmm4
; SSE2-NEXT: cvttps2dq %xmm4, %xmm4
; SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm3[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm3
; SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm6, %xmm4
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
; SSE2-NEXT: psrad $16, %xmm3
; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3]
; SSE2-NEXT: pslld $23, %xmm2
; SSE2-NEXT: paddd %xmm5, %xmm2
; SSE2-NEXT: cvttps2dq %xmm2, %xmm2
; SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm2, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
; SSE2-NEXT: pmuludq %xmm4, %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE2-NEXT: psrad $16, %xmm0
; SSE2-NEXT: packssdw %xmm3, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_funnnel_v8i16:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm1, %xmm3
; SSE41-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE41-NEXT: pmovzxwd {{.*#+}} xmm4 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
; SSE41-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4,4,5,5,6,6,7,7]
; SSE41-NEXT: pslld $23, %xmm2
; SSE41-NEXT: movdqa {{.*#+}} xmm5 = [1065353216,1065353216,1065353216,1065353216]
; SSE41-NEXT: paddd %xmm5, %xmm2
; SSE41-NEXT: cvttps2dq %xmm2, %xmm2
; SSE41-NEXT: pmulld %xmm3, %xmm2
; SSE41-NEXT: psrld $16, %xmm2
; SSE41-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; SSE41-NEXT: pslld $23, %xmm4
; SSE41-NEXT: paddd %xmm5, %xmm4
; SSE41-NEXT: cvttps2dq %xmm4, %xmm0
; SSE41-NEXT: pmulld %xmm1, %xmm0
; SSE41-NEXT: psrld $16, %xmm0
; SSE41-NEXT: packusdw %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_funnnel_v8i16:
; AVX1: # %bb.0:
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm3 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm4 = xmm2[4,4,5,5,6,6,7,7]
; AVX1-NEXT: vpslld $23, %xmm4, %xmm4
; AVX1-NEXT: vbroadcastss {{.*#+}} xmm5 = [1065353216,1065353216,1065353216,1065353216]
; AVX1-NEXT: vpaddd %xmm5, %xmm4, %xmm4
; AVX1-NEXT: vcvttps2dq %xmm4, %xmm4
; AVX1-NEXT: vpmulld %xmm4, %xmm3, %xmm3
; AVX1-NEXT: vpsrld $16, %xmm3, %xmm3
; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
; AVX1-NEXT: vpslld $23, %xmm1, %xmm1
; AVX1-NEXT: vpaddd %xmm5, %xmm1, %xmm1
; AVX1-NEXT: vcvttps2dq %xmm1, %xmm1
; AVX1-NEXT: vpmulld %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpsrld $16, %xmm0, %xmm0
; AVX1-NEXT: vpackusdw %xmm3, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_funnnel_v8i16:
; AVX2: # %bb.0:
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; AVX2-NEXT: vpslld $16, %ymm0, %ymm0
; AVX2-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm1
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX2-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT: vpackusdw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: var_funnnel_v8i16:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX512F-NEXT: vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; AVX512F-NEXT: vpslld $16, %ymm0, %ymm0
; AVX512F-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
; AVX512F-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm1
; AVX512F-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX512F-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX512F-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512F-NEXT: vpmovdw %zmm0, %ymm0
; AVX512F-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: var_funnnel_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX512VL-NEXT: vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; AVX512VL-NEXT: vpslld $16, %ymm0, %ymm0
; AVX512VL-NEXT: vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
; AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm1
; AVX512VL-NEXT: vpmovzxwd {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
; AVX512VL-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX512VL-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512VL-NEXT: vpmovdw %ymm0, %xmm0
; AVX512VL-NEXT: vzeroupper
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: var_funnnel_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT: vpbroadcastw {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15]
; AVX512BW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512BW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlvw %zmm4, %zmm1, %zmm1
; AVX512BW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512BW-NEXT: vpsllvw %zmm2, %zmm0, %zmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: var_funnnel_v8i16:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm2 killed $xmm2 def $zmm2
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldvw %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: var_funnnel_v8i16:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpbroadcastw {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15]
; AVX512VLBW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VLBW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlvw %xmm4, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllvw %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: var_funnnel_v8i16:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvw %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: var_funnnel_v8i16:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15]
; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm4
; XOPAVX1-NEXT: vpshlw %xmm4, %xmm0, %xmm0
; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm2
; XOPAVX1-NEXT: vpxor %xmm3, %xmm3, %xmm3
; XOPAVX1-NEXT: vpsubw %xmm2, %xmm3, %xmm2
; XOPAVX1-NEXT: vpsrlw $1, %xmm1, %xmm1
; XOPAVX1-NEXT: vpshlw %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_funnnel_v8i16:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpbroadcastw {{.*#+}} xmm3 = [15,15,15,15,15,15,15,15]
; XOPAVX2-NEXT: vpand %xmm3, %xmm2, %xmm4
; XOPAVX2-NEXT: vpshlw %xmm4, %xmm0, %xmm0
; XOPAVX2-NEXT: vpandn %xmm3, %xmm2, %xmm2
; XOPAVX2-NEXT: vpxor %xmm3, %xmm3, %xmm3
; XOPAVX2-NEXT: vpsubw %xmm2, %xmm3, %xmm2
; XOPAVX2-NEXT: vpsrlw $1, %xmm1, %xmm1
; XOPAVX2-NEXT: vpshlw %xmm2, %xmm1, %xmm1
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: var_funnnel_v8i16:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: movdqa %xmm2, %xmm5
; X86-SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4,4,5,5,6,6,7,7]
; X86-SSE2-NEXT: pslld $23, %xmm5
; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [1065353216,1065353216,1065353216,1065353216]
; X86-SSE2-NEXT: paddd %xmm4, %xmm5
; X86-SSE2-NEXT: cvttps2dq %xmm5, %xmm5
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm6 = xmm3[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm5, %xmm3
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm6, %xmm5
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
; X86-SSE2-NEXT: psrad $16, %xmm3
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3]
; X86-SSE2-NEXT: pslld $23, %xmm2
; X86-SSE2-NEXT: paddd %xmm4, %xmm2
; X86-SSE2-NEXT: cvttps2dq %xmm2, %xmm2
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm2, %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
; X86-SSE2-NEXT: pmuludq %xmm4, %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE2-NEXT: psrad $16, %xmm0
; X86-SSE2-NEXT: packssdw %xmm3, %xmm0
; X86-SSE2-NEXT: retl
%res = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> %amt)
ret <8 x i16> %res
}
define <16 x i8> @var_funnnel_v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> %amt) nounwind {
; SSE2-LABEL: var_funnnel_v16i8:
; SSE2: # %bb.0:
; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE2-NEXT: pxor %xmm5, %xmm5
; SSE2-NEXT: movdqa %xmm2, %xmm4
; SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; SSE2-NEXT: movdqa %xmm4, %xmm6
; SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4,4,5,5,6,6,7,7]
; SSE2-NEXT: pslld $23, %xmm6
; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; SSE2-NEXT: paddd %xmm3, %xmm6
; SSE2-NEXT: cvttps2dq %xmm6, %xmm6
; SSE2-NEXT: pslld $16, %xmm6
; SSE2-NEXT: psrad $16, %xmm6
; SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0,0,1,1,2,2,3,3]
; SSE2-NEXT: pslld $23, %xmm4
; SSE2-NEXT: paddd %xmm3, %xmm4
; SSE2-NEXT: cvttps2dq %xmm4, %xmm7
; SSE2-NEXT: pslld $16, %xmm7
; SSE2-NEXT: psrad $16, %xmm7
; SSE2-NEXT: packssdw %xmm6, %xmm7
; SSE2-NEXT: movdqa %xmm1, %xmm4
; SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm0[8],xmm4[9],xmm0[9],xmm4[10],xmm0[10],xmm4[11],xmm0[11],xmm4[12],xmm0[12],xmm4[13],xmm0[13],xmm4[14],xmm0[14],xmm4[15],xmm0[15]
; SSE2-NEXT: pmullw %xmm7, %xmm4
; SSE2-NEXT: psrlw $8, %xmm4
; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; SSE2-NEXT: movdqa %xmm2, %xmm5
; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4,4,5,5,6,6,7,7]
; SSE2-NEXT: pslld $23, %xmm5
; SSE2-NEXT: paddd %xmm3, %xmm5
; SSE2-NEXT: cvttps2dq %xmm5, %xmm5
; SSE2-NEXT: pslld $16, %xmm5
; SSE2-NEXT: psrad $16, %xmm5
; SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3]
; SSE2-NEXT: pslld $23, %xmm2
; SSE2-NEXT: paddd %xmm3, %xmm2
; SSE2-NEXT: cvttps2dq %xmm2, %xmm2
; SSE2-NEXT: pslld $16, %xmm2
; SSE2-NEXT: psrad $16, %xmm2
; SSE2-NEXT: packssdw %xmm5, %xmm2
; SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE2-NEXT: pmullw %xmm1, %xmm2
; SSE2-NEXT: psrlw $8, %xmm2
; SSE2-NEXT: packuswb %xmm4, %xmm2
; SSE2-NEXT: movdqa %xmm2, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_funnnel_v16i8:
; SSE41: # %bb.0:
; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE41-NEXT: pxor %xmm3, %xmm3
; SSE41-NEXT: pmovzxbd {{.*#+}} xmm4 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
; SSE41-NEXT: pmovzxbw {{.*#+}} xmm5 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
; SSE41-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm3[8],xmm2[9],xmm3[9],xmm2[10],xmm3[10],xmm2[11],xmm3[11],xmm2[12],xmm3[12],xmm2[13],xmm3[13],xmm2[14],xmm3[14],xmm2[15],xmm3[15]
; SSE41-NEXT: pmovzxwd {{.*#+}} xmm3 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
; SSE41-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4,4,5,5,6,6,7,7]
; SSE41-NEXT: pslld $23, %xmm2
; SSE41-NEXT: movdqa {{.*#+}} xmm6 = [1065353216,1065353216,1065353216,1065353216]
; SSE41-NEXT: paddd %xmm6, %xmm2
; SSE41-NEXT: cvttps2dq %xmm2, %xmm2
; SSE41-NEXT: pslld $23, %xmm3
; SSE41-NEXT: paddd %xmm6, %xmm3
; SSE41-NEXT: cvttps2dq %xmm3, %xmm3
; SSE41-NEXT: packusdw %xmm2, %xmm3
; SSE41-NEXT: movdqa %xmm1, %xmm7
; SSE41-NEXT: punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm0[8],xmm7[9],xmm0[9],xmm7[10],xmm0[10],xmm7[11],xmm0[11],xmm7[12],xmm0[12],xmm7[13],xmm0[13],xmm7[14],xmm0[14],xmm7[15],xmm0[15]
; SSE41-NEXT: pmullw %xmm3, %xmm7
; SSE41-NEXT: psrlw $8, %xmm7
; SSE41-NEXT: pslld $23, %xmm4
; SSE41-NEXT: paddd %xmm6, %xmm4
; SSE41-NEXT: cvttps2dq %xmm4, %xmm2
; SSE41-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4,4,5,5,6,6,7,7]
; SSE41-NEXT: pslld $23, %xmm5
; SSE41-NEXT: paddd %xmm6, %xmm5
; SSE41-NEXT: cvttps2dq %xmm5, %xmm3
; SSE41-NEXT: packusdw %xmm3, %xmm2
; SSE41-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE41-NEXT: pmullw %xmm1, %xmm2
; SSE41-NEXT: psrlw $8, %xmm2
; SSE41-NEXT: packuswb %xmm7, %xmm2
; SSE41-NEXT: movdqa %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_funnnel_v16i8:
; AVX1: # %bb.0:
; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX1-NEXT: vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm2[8],xmm3[8],xmm2[9],xmm3[9],xmm2[10],xmm3[10],xmm2[11],xmm3[11],xmm2[12],xmm3[12],xmm2[13],xmm3[13],xmm2[14],xmm3[14],xmm2[15],xmm3[15]
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm4 = xmm3[4,4,5,5,6,6,7,7]
; AVX1-NEXT: vpslld $23, %xmm4, %xmm4
; AVX1-NEXT: vbroadcastss {{.*#+}} xmm5 = [1065353216,1065353216,1065353216,1065353216]
; AVX1-NEXT: vpaddd %xmm5, %xmm4, %xmm4
; AVX1-NEXT: vcvttps2dq %xmm4, %xmm4
; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
; AVX1-NEXT: vpslld $23, %xmm3, %xmm3
; AVX1-NEXT: vpaddd %xmm5, %xmm3, %xmm3
; AVX1-NEXT: vcvttps2dq %xmm3, %xmm3
; AVX1-NEXT: vpackusdw %xmm4, %xmm3, %xmm3
; AVX1-NEXT: vpunpckhbw {{.*#+}} xmm4 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX1-NEXT: vpmullw %xmm3, %xmm4, %xmm3
; AVX1-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX1-NEXT: vpmovzxbd {{.*#+}} xmm4 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero
; AVX1-NEXT: vpslld $23, %xmm4, %xmm4
; AVX1-NEXT: vpaddd %xmm5, %xmm4, %xmm4
; AVX1-NEXT: vcvttps2dq %xmm4, %xmm4
; AVX1-NEXT: vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm2[4,4,5,5,6,6,7,7]
; AVX1-NEXT: vpslld $23, %xmm2, %xmm2
; AVX1-NEXT: vpaddd %xmm5, %xmm2, %xmm2
; AVX1-NEXT: vcvttps2dq %xmm2, %xmm2
; AVX1-NEXT: vpackusdw %xmm2, %xmm4, %xmm2
; AVX1-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX1-NEXT: vpmullw %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX1-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_funnnel_v16i8:
; AVX2: # %bb.0:
; AVX2-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX2-NEXT: vpxor %xmm4, %xmm4, %xmm4
; AVX2-NEXT: vpunpckhbw {{.*#+}} xmm4 = xmm2[8],xmm4[8],xmm2[9],xmm4[9],xmm2[10],xmm4[10],xmm2[11],xmm4[11],xmm2[12],xmm4[12],xmm2[13],xmm4[13],xmm2[14],xmm4[14],xmm2[15],xmm4[15]
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
; AVX2-NEXT: vpsllvd %ymm4, %ymm3, %ymm3
; AVX2-NEXT: vmovdqa {{.*#+}} ymm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
; AVX2-NEXT: vpshufb %ymm4, %ymm3, %ymm3
; AVX2-NEXT: vpermq {{.*#+}} ymm3 = ymm3[0,2,2,3]
; AVX2-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX2-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; AVX2-NEXT: vpmovzxbd {{.*#+}} ymm1 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero
; AVX2-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vpshufb %ymm4, %ymm0, %ymm0
; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX2-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512F-LABEL: var_funnnel_v16i8:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpbroadcastb {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
; AVX512F-NEXT: vpand %xmm3, %xmm2, %xmm4
; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero,xmm4[4],zero,zero,zero,xmm4[5],zero,zero,zero,xmm4[6],zero,zero,zero,xmm4[7],zero,zero,zero,xmm4[8],zero,zero,zero,xmm4[9],zero,zero,zero,xmm4[10],zero,zero,zero,xmm4[11],zero,zero,zero,xmm4[12],zero,zero,zero,xmm4[13],zero,zero,zero,xmm4[14],zero,zero,zero,xmm4[15],zero,zero,zero
; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512F-NEXT: vpsllvd %zmm4, %zmm0, %zmm0
; AVX512F-NEXT: vpandn %xmm3, %xmm2, %xmm2
; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
; AVX512F-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512F-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512F-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512F-NEXT: vpsrlvd %zmm2, %zmm1, %zmm1
; AVX512F-NEXT: vpord %zmm1, %zmm0, %zmm0
; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: var_funnnel_v16i8:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpbroadcastd {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
; AVX512VL-NEXT: vpand %xmm3, %xmm2, %xmm4
; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm4 = xmm4[0],zero,zero,zero,xmm4[1],zero,zero,zero,xmm4[2],zero,zero,zero,xmm4[3],zero,zero,zero,xmm4[4],zero,zero,zero,xmm4[5],zero,zero,zero,xmm4[6],zero,zero,zero,xmm4[7],zero,zero,zero,xmm4[8],zero,zero,zero,xmm4[9],zero,zero,zero,xmm4[10],zero,zero,zero,xmm4[11],zero,zero,zero,xmm4[12],zero,zero,zero,xmm4[13],zero,zero,zero,xmm4[14],zero,zero,zero,xmm4[15],zero,zero,zero
; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; AVX512VL-NEXT: vpsllvd %zmm4, %zmm0, %zmm0
; AVX512VL-NEXT: vpandn %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm2 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero,xmm2[2],zero,zero,zero,xmm2[3],zero,zero,zero,xmm2[4],zero,zero,zero,xmm2[5],zero,zero,zero,xmm2[6],zero,zero,zero,xmm2[7],zero,zero,zero,xmm2[8],zero,zero,zero,xmm2[9],zero,zero,zero,xmm2[10],zero,zero,zero,xmm2[11],zero,zero,zero,xmm2[12],zero,zero,zero,xmm2[13],zero,zero,zero,xmm2[14],zero,zero,zero,xmm2[15],zero,zero,zero
; AVX512VL-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm1, %xmm1
; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero
; AVX512VL-NEXT: vpsrlvd %zmm2, %zmm1, %zmm1
; AVX512VL-NEXT: vpord %zmm1, %zmm0, %zmm0
; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0
; AVX512VL-NEXT: vzeroupper
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: var_funnnel_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero,xmm2[8],zero,xmm2[9],zero,xmm2[10],zero,xmm2[11],zero,xmm2[12],zero,xmm2[13],zero,xmm2[14],zero,xmm2[15],zero
; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512BW-NEXT: vpsllw $8, %ymm0, %ymm0
; AVX512BW-NEXT: vpor %ymm1, %ymm0, %ymm0
; AVX512BW-NEXT: vpsllvw %zmm2, %zmm0, %zmm0
; AVX512BW-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: var_funnnel_v16i8:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vmovdqa {{.*#+}} ymm3 = [0,64,1,65,2,66,3,67,4,68,5,69,6,70,7,71,8,72,9,73,10,74,11,75,12,76,13,77,14,78,15,79]
; AVX512VBMI2-NEXT: vpermt2b %zmm0, %zmm3, %zmm1
; AVX512VBMI2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm0
; AVX512VBMI2-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512VBMI2-NEXT: vpsllvw %zmm0, %zmm1, %zmm0
; AVX512VBMI2-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VBMI2-NEXT: vpmovwb %zmm0, %ymm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: var_funnnel_v16i8:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
; AVX512VLBW-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512VLBW-NEXT: vpsllw $8, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpor %ymm1, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm1
; AVX512VLBW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
; AVX512VLBW-NEXT: vpsllvw %ymm1, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpmovwb %ymm0, %xmm0
; AVX512VLBW-NEXT: vzeroupper
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: var_funnnel_v16i8:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $ymm1
; AVX512VLVBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512VLVBMI2-NEXT: vmovdqa {{.*#+}} ymm3 = [0,32,1,33,2,34,3,35,4,36,5,37,6,38,7,39,8,40,9,41,10,42,11,43,12,44,13,45,14,46,15,47]
; AVX512VLVBMI2-NEXT: vpermi2b %ymm0, %ymm1, %ymm3
; AVX512VLVBMI2-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm0
; AVX512VLVBMI2-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512VLVBMI2-NEXT: vpsllvw %ymm0, %ymm3, %ymm0
; AVX512VLVBMI2-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VLVBMI2-NEXT: vpmovwb %ymm0, %xmm0
; AVX512VLVBMI2-NEXT: vzeroupper
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: var_funnnel_v16i8:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpcmpeqd %xmm3, %xmm3, %xmm3
; XOPAVX1-NEXT: vpshlb %xmm3, %xmm1, %xmm1
; XOPAVX1-NEXT: vbroadcastss {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
; XOPAVX1-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOPAVX1-NEXT: vpxor %xmm5, %xmm5, %xmm5
; XOPAVX1-NEXT: vpsubb %xmm4, %xmm5, %xmm4
; XOPAVX1-NEXT: vpshlb %xmm4, %xmm1, %xmm1
; XOPAVX1-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOPAVX1-NEXT: vpshlb %xmm2, %xmm0, %xmm0
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_funnnel_v16i8:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpcmpeqd %xmm3, %xmm3, %xmm3
; XOPAVX2-NEXT: vpshlb %xmm3, %xmm1, %xmm1
; XOPAVX2-NEXT: vpbroadcastb {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
; XOPAVX2-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOPAVX2-NEXT: vpxor %xmm5, %xmm5, %xmm5
; XOPAVX2-NEXT: vpsubb %xmm4, %xmm5, %xmm4
; XOPAVX2-NEXT: vpshlb %xmm4, %xmm1, %xmm1
; XOPAVX2-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOPAVX2-NEXT: vpshlb %xmm2, %xmm0, %xmm0
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: var_funnnel_v16i8:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: pxor %xmm5, %xmm5
; X86-SSE2-NEXT: movdqa %xmm2, %xmm4
; X86-SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm5[8],xmm4[9],xmm5[9],xmm4[10],xmm5[10],xmm4[11],xmm5[11],xmm4[12],xmm5[12],xmm4[13],xmm5[13],xmm4[14],xmm5[14],xmm4[15],xmm5[15]
; X86-SSE2-NEXT: movdqa %xmm4, %xmm6
; X86-SSE2-NEXT: punpckhwd {{.*#+}} xmm6 = xmm6[4,4,5,5,6,6,7,7]
; X86-SSE2-NEXT: pslld $23, %xmm6
; X86-SSE2-NEXT: movdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; X86-SSE2-NEXT: paddd %xmm3, %xmm6
; X86-SSE2-NEXT: cvttps2dq %xmm6, %xmm6
; X86-SSE2-NEXT: pslld $16, %xmm6
; X86-SSE2-NEXT: psrad $16, %xmm6
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0,0,1,1,2,2,3,3]
; X86-SSE2-NEXT: pslld $23, %xmm4
; X86-SSE2-NEXT: paddd %xmm3, %xmm4
; X86-SSE2-NEXT: cvttps2dq %xmm4, %xmm7
; X86-SSE2-NEXT: pslld $16, %xmm7
; X86-SSE2-NEXT: psrad $16, %xmm7
; X86-SSE2-NEXT: packssdw %xmm6, %xmm7
; X86-SSE2-NEXT: movdqa %xmm1, %xmm4
; X86-SSE2-NEXT: punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm0[8],xmm4[9],xmm0[9],xmm4[10],xmm0[10],xmm4[11],xmm0[11],xmm4[12],xmm0[12],xmm4[13],xmm0[13],xmm4[14],xmm0[14],xmm4[15],xmm0[15]
; X86-SSE2-NEXT: pmullw %xmm7, %xmm4
; X86-SSE2-NEXT: psrlw $8, %xmm4
; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1],xmm2[2],xmm5[2],xmm2[3],xmm5[3],xmm2[4],xmm5[4],xmm2[5],xmm5[5],xmm2[6],xmm5[6],xmm2[7],xmm5[7]
; X86-SSE2-NEXT: movdqa %xmm2, %xmm5
; X86-SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4,4,5,5,6,6,7,7]
; X86-SSE2-NEXT: pslld $23, %xmm5
; X86-SSE2-NEXT: paddd %xmm3, %xmm5
; X86-SSE2-NEXT: cvttps2dq %xmm5, %xmm5
; X86-SSE2-NEXT: pslld $16, %xmm5
; X86-SSE2-NEXT: psrad $16, %xmm5
; X86-SSE2-NEXT: punpcklwd {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3]
; X86-SSE2-NEXT: pslld $23, %xmm2
; X86-SSE2-NEXT: paddd %xmm3, %xmm2
; X86-SSE2-NEXT: cvttps2dq %xmm2, %xmm2
; X86-SSE2-NEXT: pslld $16, %xmm2
; X86-SSE2-NEXT: psrad $16, %xmm2
; X86-SSE2-NEXT: packssdw %xmm5, %xmm2
; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; X86-SSE2-NEXT: pmullw %xmm1, %xmm2
; X86-SSE2-NEXT: psrlw $8, %xmm2
; X86-SSE2-NEXT: packuswb %xmm4, %xmm2
; X86-SSE2-NEXT: movdqa %xmm2, %xmm0
; X86-SSE2-NEXT: retl
%res = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> %amt)
ret <16 x i8> %res
}
;
; Uniform Variable Shifts
;
define <2 x i64> @splatvar_funnnel_v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> %amt) nounwind {
; SSE2-LABEL: splatvar_funnnel_v2i64:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [63,63]
; SSE2-NEXT: movdqa %xmm2, %xmm4
; SSE2-NEXT: pandn %xmm3, %xmm4
; SSE2-NEXT: psrlq $1, %xmm1
; SSE2-NEXT: psrlq %xmm4, %xmm1
; SSE2-NEXT: pand %xmm3, %xmm2
; SSE2-NEXT: psllq %xmm2, %xmm0
; SSE2-NEXT: por %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_funnnel_v2i64:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovsxbq {{.*#+}} xmm3 = [63,63]
; SSE41-NEXT: movdqa %xmm2, %xmm4
; SSE41-NEXT: pandn %xmm3, %xmm4
; SSE41-NEXT: psrlq $1, %xmm1
; SSE41-NEXT: psrlq %xmm4, %xmm1
; SSE41-NEXT: pand %xmm3, %xmm2
; SSE41-NEXT: psllq %xmm2, %xmm0
; SSE41-NEXT: por %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: splatvar_funnnel_v2i64:
; AVX: # %bb.0:
; AVX-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatvar_funnnel_v2i64:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX512F-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512F-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512F-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX512F-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512F-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatvar_funnnel_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpbroadcastq {{.*#+}} xmm3 = [63,63]
; AVX512VL-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VL-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX512VL-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatvar_funnnel_v2i64:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; AVX512BW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512BW-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX512BW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512BW-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatvar_funnnel_v2i64:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpbroadcastq %xmm2, %xmm2
; AVX512VBMI2-NEXT: vpshldvq %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatvar_funnnel_v2i64:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpbroadcastq {{.*#+}} xmm3 = [63,63]
; AVX512VLBW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VLBW-NEXT: vpsrlq $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatvar_funnnel_v2i64:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpbroadcastq %xmm2, %xmm2
; AVX512VLVBMI2-NEXT: vpshldvq %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatvar_funnnel_v2i64:
; XOP: # %bb.0:
; XOP-NEXT: vpmovsxbq {{.*#+}} xmm3 = [63,63]
; XOP-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOP-NEXT: vpsrlq $1, %xmm1, %xmm1
; XOP-NEXT: vpsrlq %xmm4, %xmm1, %xmm1
; XOP-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOP-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatvar_funnnel_v2i64:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa {{.*#+}} xmm3 = [63,0,63,0]
; X86-SSE2-NEXT: movdqa %xmm2, %xmm4
; X86-SSE2-NEXT: pandn %xmm3, %xmm4
; X86-SSE2-NEXT: psrlq $1, %xmm1
; X86-SSE2-NEXT: psrlq %xmm4, %xmm1
; X86-SSE2-NEXT: pand %xmm3, %xmm2
; X86-SSE2-NEXT: psllq %xmm2, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%splat = shufflevector <2 x i64> %amt, <2 x i64> undef, <2 x i32> zeroinitializer
%res = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> %splat)
ret <2 x i64> %res
}
define <4 x i32> @splatvar_funnnel_v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> %amt) nounwind {
; SSE-LABEL: splatvar_funnnel_v4i32:
; SSE: # %bb.0:
; SSE-NEXT: movdqa %xmm1, %xmm3
; SSE-NEXT: punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE-NEXT: psllq %xmm2, %xmm3
; SSE-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-NEXT: psllq %xmm2, %xmm1
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,3],xmm3[1,3]
; SSE-NEXT: movaps %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_funnnel_v4i32:
; AVX: # %bb.0:
; AVX-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; AVX-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatvar_funnnel_v4i32:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX512F-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; AVX512F-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX512F-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatvar_funnnel_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX512VL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX512VL-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatvar_funnnel_v4i32:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX512BW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512BW-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; AVX512BW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX512BW-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatvar_funnnel_v4i32:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpbroadcastd %xmm2, %xmm2
; AVX512VBMI2-NEXT: vpshldvd %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatvar_funnnel_v4i32:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; AVX512VLBW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; AVX512VLBW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; AVX512VLBW-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatvar_funnnel_v4i32:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpbroadcastd %xmm2, %xmm2
; AVX512VLVBMI2-NEXT: vpshldvd %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatvar_funnnel_v4i32:
; XOP: # %bb.0:
; XOP-NEXT: vpunpckhdq {{.*#+}} xmm3 = xmm1[2],xmm0[2],xmm1[3],xmm0[3]
; XOP-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; XOP-NEXT: vpsllq %xmm2, %xmm3, %xmm3
; XOP-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; XOP-NEXT: vpsllq %xmm2, %xmm0, %xmm0
; XOP-NEXT: vshufps {{.*#+}} xmm0 = xmm0[1,3],xmm3[1,3]
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatvar_funnnel_v4i32:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: punpckhdq {{.*#+}} xmm3 = xmm3[2],xmm0[2],xmm3[3],xmm0[3]
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: psllq %xmm2, %xmm3
; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; X86-SSE2-NEXT: psllq %xmm2, %xmm1
; X86-SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,3],xmm3[1,3]
; X86-SSE2-NEXT: movaps %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%splat = shufflevector <4 x i32> %amt, <4 x i32> undef, <4 x i32> zeroinitializer
%res = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> %splat)
ret <4 x i32> %res
}
define <8 x i16> @splatvar_funnnel_v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> %amt) nounwind {
; SSE2-LABEL: splatvar_funnnel_v8i16:
; SSE2: # %bb.0:
; SSE2-NEXT: movd {{.*#+}} xmm3 = [15,0,0,0]
; SSE2-NEXT: movdqa %xmm2, %xmm4
; SSE2-NEXT: pandn %xmm3, %xmm4
; SSE2-NEXT: psrlw $1, %xmm1
; SSE2-NEXT: psrlw %xmm4, %xmm1
; SSE2-NEXT: pand %xmm3, %xmm2
; SSE2-NEXT: psllw %xmm2, %xmm0
; SSE2-NEXT: por %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_funnnel_v8i16:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovsxbq {{.*#+}} xmm3 = [15,0]
; SSE41-NEXT: movdqa %xmm2, %xmm4
; SSE41-NEXT: pandn %xmm3, %xmm4
; SSE41-NEXT: psrlw $1, %xmm1
; SSE41-NEXT: psrlw %xmm4, %xmm1
; SSE41-NEXT: pand %xmm3, %xmm2
; SSE41-NEXT: psllw %xmm2, %xmm0
; SSE41-NEXT: por %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: splatvar_funnnel_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; AVX-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; AVX-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatvar_funnnel_v8i16:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; AVX512F-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512F-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512F-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; AVX512F-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512F-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatvar_funnnel_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; AVX512VL-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VL-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; AVX512VL-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatvar_funnnel_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; AVX512BW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512BW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; AVX512BW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512BW-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatvar_funnnel_v8i16:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpbroadcastw %xmm2, %xmm2
; AVX512VBMI2-NEXT: vpshldvw %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatvar_funnnel_v8i16:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; AVX512VLBW-NEXT: vpandn %xmm3, %xmm2, %xmm4
; AVX512VLBW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpand %xmm3, %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatvar_funnnel_v8i16:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpbroadcastw %xmm2, %xmm2
; AVX512VLVBMI2-NEXT: vpshldvw %xmm2, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatvar_funnnel_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpmovsxbq {{.*#+}} xmm3 = [15,0]
; XOP-NEXT: vpandn %xmm3, %xmm2, %xmm4
; XOP-NEXT: vpsrlw $1, %xmm1, %xmm1
; XOP-NEXT: vpsrlw %xmm4, %xmm1, %xmm1
; XOP-NEXT: vpand %xmm3, %xmm2, %xmm2
; XOP-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatvar_funnnel_v8i16:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movd {{.*#+}} xmm3 = [15,0,0,0]
; X86-SSE2-NEXT: movdqa %xmm2, %xmm4
; X86-SSE2-NEXT: pandn %xmm3, %xmm4
; X86-SSE2-NEXT: psrlw $1, %xmm1
; X86-SSE2-NEXT: psrlw %xmm4, %xmm1
; X86-SSE2-NEXT: pand %xmm3, %xmm2
; X86-SSE2-NEXT: psllw %xmm2, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%splat = shufflevector <8 x i16> %amt, <8 x i16> undef, <8 x i32> zeroinitializer
%res = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> %splat)
ret <8 x i16> %res
}
define <16 x i8> @splatvar_funnnel_v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> %amt) nounwind {
; SSE-LABEL: splatvar_funnnel_v16i8:
; SSE: # %bb.0:
; SSE-NEXT: movdqa %xmm1, %xmm3
; SSE-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm0[8],xmm3[9],xmm0[9],xmm3[10],xmm0[10],xmm3[11],xmm0[11],xmm3[12],xmm0[12],xmm3[13],xmm0[13],xmm3[14],xmm0[14],xmm3[15],xmm0[15]
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE-NEXT: psllw %xmm2, %xmm3
; SSE-NEXT: psrlw $8, %xmm3
; SSE-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE-NEXT: psllw %xmm2, %xmm1
; SSE-NEXT: psrlw $8, %xmm1
; SSE-NEXT: packuswb %xmm3, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_funnnel_v16i8:
; AVX: # %bb.0:
; AVX-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatvar_funnnel_v16i8:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512F-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512F-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512F-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512F-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512F-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatvar_funnnel_v16i8:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512VL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512VL-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512VL-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512VL-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512VL-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatvar_funnnel_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512BW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512BW-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512BW-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512BW-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512BW-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512BW-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatvar_funnnel_v16i8:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512VBMI2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VBMI2-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512VBMI2-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512VBMI2-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512VBMI2-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VBMI2-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512VBMI2-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatvar_funnnel_v16i8:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512VLBW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VLBW-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512VLBW-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512VLBW-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512VLBW-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatvar_funnnel_v16i8:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512VLVBMI2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX512VLVBMI2-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; AVX512VLVBMI2-NEXT: vpsrlw $8, %xmm3, %xmm3
; AVX512VLVBMI2-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512VLVBMI2-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: vpackuswb %xmm3, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatvar_funnnel_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; XOP-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; XOP-NEXT: vpsllw %xmm2, %xmm3, %xmm3
; XOP-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; XOP-NEXT: vpsllw %xmm2, %xmm0, %xmm0
; XOP-NEXT: vpperm {{.*#+}} xmm0 = xmm0[1,3,5,7,9,11,13,15],xmm3[1,3,5,7,9,11,13,15]
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatvar_funnnel_v16i8:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm0[8],xmm3[9],xmm0[9],xmm3[10],xmm0[10],xmm3[11],xmm0[11],xmm3[12],xmm0[12],xmm3[13],xmm0[13],xmm3[14],xmm0[14],xmm3[15],xmm0[15]
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: psllw %xmm2, %xmm3
; X86-SSE2-NEXT: psrlw $8, %xmm3
; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; X86-SSE2-NEXT: psllw %xmm2, %xmm1
; X86-SSE2-NEXT: psrlw $8, %xmm1
; X86-SSE2-NEXT: packuswb %xmm3, %xmm1
; X86-SSE2-NEXT: movdqa %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%splat = shufflevector <16 x i8> %amt, <16 x i8> undef, <16 x i32> zeroinitializer
%res = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> %splat)
ret <16 x i8> %res
}
; CGP should allow a cross-block splat shift amount to be seen in SDAG.
; PR37426 - https://bugs.llvm.org/show_bug.cgi?id=37426
define void @sink_splatvar(ptr %p, i32 %shift_amt) {
; SSE-LABEL: sink_splatvar:
; SSE: # %bb.0: # %entry
; SSE-NEXT: movd %esi, %xmm0
; SSE-NEXT: movq $-1024, %rax # imm = 0xFC00
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: .p2align 4, 0x90
; SSE-NEXT: .LBB8_1: # %loop
; SSE-NEXT: # =>This Inner Loop Header: Depth=1
; SSE-NEXT: movdqu 1024(%rdi,%rax), %xmm1
; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,2,3,3]
; SSE-NEXT: psllq %xmm0, %xmm2
; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; SSE-NEXT: psllq %xmm0, %xmm1
; SSE-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
; SSE-NEXT: movups %xmm1, 1024(%rdi,%rax)
; SSE-NEXT: addq $16, %rax
; SSE-NEXT: jne .LBB8_1
; SSE-NEXT: # %bb.2: # %end
; SSE-NEXT: retq
;
; AVX1-LABEL: sink_splatvar:
; AVX1: # %bb.0: # %entry
; AVX1-NEXT: vmovd %esi, %xmm0
; AVX1-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: .p2align 4, 0x90
; AVX1-NEXT: .LBB8_1: # %loop
; AVX1-NEXT: # =>This Inner Loop Header: Depth=1
; AVX1-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX1-NEXT: vpshufd {{.*#+}} xmm2 = xmm1[2,2,3,3]
; AVX1-NEXT: vpsllq %xmm0, %xmm2, %xmm2
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; AVX1-NEXT: vpsllq %xmm0, %xmm1, %xmm1
; AVX1-NEXT: vshufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
; AVX1-NEXT: vmovups %xmm1, 1024(%rdi,%rax)
; AVX1-NEXT: addq $16, %rax
; AVX1-NEXT: jne .LBB8_1
; AVX1-NEXT: # %bb.2: # %end
; AVX1-NEXT: retq
;
; AVX2-LABEL: sink_splatvar:
; AVX2: # %bb.0: # %entry
; AVX2-NEXT: vmovd %esi, %xmm0
; AVX2-NEXT: vpbroadcastd %xmm0, %xmm0
; AVX2-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX2-NEXT: vpbroadcastd {{.*#+}} xmm1 = [31,31,31,31]
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vpbroadcastd {{.*#+}} xmm1 = [32,32,32,32]
; AVX2-NEXT: vpsubd %xmm0, %xmm1, %xmm1
; AVX2-NEXT: .p2align 4, 0x90
; AVX2-NEXT: .LBB8_1: # %loop
; AVX2-NEXT: # =>This Inner Loop Header: Depth=1
; AVX2-NEXT: vmovdqu 1024(%rdi,%rax), %xmm2
; AVX2-NEXT: vpsllvd %xmm0, %xmm2, %xmm3
; AVX2-NEXT: vpsrlvd %xmm1, %xmm2, %xmm2
; AVX2-NEXT: vpor %xmm2, %xmm3, %xmm2
; AVX2-NEXT: vmovdqu %xmm2, 1024(%rdi,%rax)
; AVX2-NEXT: addq $16, %rax
; AVX2-NEXT: jne .LBB8_1
; AVX2-NEXT: # %bb.2: # %end
; AVX2-NEXT: retq
;
; AVX512F-LABEL: sink_splatvar:
; AVX512F: # %bb.0: # %entry
; AVX512F-NEXT: vmovd %esi, %xmm0
; AVX512F-NEXT: vpbroadcastd %xmm0, %xmm0
; AVX512F-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512F-NEXT: .p2align 4, 0x90
; AVX512F-NEXT: .LBB8_1: # %loop
; AVX512F-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512F-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512F-NEXT: vprolvd %zmm0, %zmm1, %zmm1
; AVX512F-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512F-NEXT: addq $16, %rax
; AVX512F-NEXT: jne .LBB8_1
; AVX512F-NEXT: # %bb.2: # %end
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: sink_splatvar:
; AVX512VL: # %bb.0: # %entry
; AVX512VL-NEXT: vpbroadcastd %esi, %xmm0
; AVX512VL-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512VL-NEXT: .p2align 4, 0x90
; AVX512VL-NEXT: .LBB8_1: # %loop
; AVX512VL-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512VL-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512VL-NEXT: vprolvd %xmm0, %xmm1, %xmm1
; AVX512VL-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512VL-NEXT: addq $16, %rax
; AVX512VL-NEXT: jne .LBB8_1
; AVX512VL-NEXT: # %bb.2: # %end
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: sink_splatvar:
; AVX512BW: # %bb.0: # %entry
; AVX512BW-NEXT: vmovd %esi, %xmm0
; AVX512BW-NEXT: vpbroadcastd %xmm0, %xmm0
; AVX512BW-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512BW-NEXT: .p2align 4, 0x90
; AVX512BW-NEXT: .LBB8_1: # %loop
; AVX512BW-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512BW-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512BW-NEXT: vprolvd %zmm0, %zmm1, %zmm1
; AVX512BW-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512BW-NEXT: addq $16, %rax
; AVX512BW-NEXT: jne .LBB8_1
; AVX512BW-NEXT: # %bb.2: # %end
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: sink_splatvar:
; AVX512VBMI2: # %bb.0: # %entry
; AVX512VBMI2-NEXT: vmovd %esi, %xmm0
; AVX512VBMI2-NEXT: vpbroadcastd %xmm0, %xmm0
; AVX512VBMI2-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512VBMI2-NEXT: .p2align 4, 0x90
; AVX512VBMI2-NEXT: .LBB8_1: # %loop
; AVX512VBMI2-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512VBMI2-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512VBMI2-NEXT: vprolvd %zmm0, %zmm1, %zmm1
; AVX512VBMI2-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512VBMI2-NEXT: addq $16, %rax
; AVX512VBMI2-NEXT: jne .LBB8_1
; AVX512VBMI2-NEXT: # %bb.2: # %end
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: sink_splatvar:
; AVX512VLBW: # %bb.0: # %entry
; AVX512VLBW-NEXT: vpbroadcastd %esi, %xmm0
; AVX512VLBW-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512VLBW-NEXT: .p2align 4, 0x90
; AVX512VLBW-NEXT: .LBB8_1: # %loop
; AVX512VLBW-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512VLBW-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512VLBW-NEXT: vprolvd %xmm0, %xmm1, %xmm1
; AVX512VLBW-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512VLBW-NEXT: addq $16, %rax
; AVX512VLBW-NEXT: jne .LBB8_1
; AVX512VLBW-NEXT: # %bb.2: # %end
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: sink_splatvar:
; AVX512VLVBMI2: # %bb.0: # %entry
; AVX512VLVBMI2-NEXT: vpbroadcastd %esi, %xmm0
; AVX512VLVBMI2-NEXT: movq $-1024, %rax # imm = 0xFC00
; AVX512VLVBMI2-NEXT: .p2align 4, 0x90
; AVX512VLVBMI2-NEXT: .LBB8_1: # %loop
; AVX512VLVBMI2-NEXT: # =>This Inner Loop Header: Depth=1
; AVX512VLVBMI2-NEXT: vmovdqu 1024(%rdi,%rax), %xmm1
; AVX512VLVBMI2-NEXT: vprolvd %xmm0, %xmm1, %xmm1
; AVX512VLVBMI2-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; AVX512VLVBMI2-NEXT: addq $16, %rax
; AVX512VLVBMI2-NEXT: jne .LBB8_1
; AVX512VLVBMI2-NEXT: # %bb.2: # %end
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: sink_splatvar:
; XOPAVX1: # %bb.0: # %entry
; XOPAVX1-NEXT: vmovd %esi, %xmm0
; XOPAVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]
; XOPAVX1-NEXT: movq $-1024, %rax # imm = 0xFC00
; XOPAVX1-NEXT: .p2align 4, 0x90
; XOPAVX1-NEXT: .LBB8_1: # %loop
; XOPAVX1-NEXT: # =>This Inner Loop Header: Depth=1
; XOPAVX1-NEXT: vprotd %xmm0, 1024(%rdi,%rax), %xmm1
; XOPAVX1-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; XOPAVX1-NEXT: addq $16, %rax
; XOPAVX1-NEXT: jne .LBB8_1
; XOPAVX1-NEXT: # %bb.2: # %end
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: sink_splatvar:
; XOPAVX2: # %bb.0: # %entry
; XOPAVX2-NEXT: vmovd %esi, %xmm0
; XOPAVX2-NEXT: vpbroadcastd %xmm0, %xmm0
; XOPAVX2-NEXT: movq $-1024, %rax # imm = 0xFC00
; XOPAVX2-NEXT: .p2align 4, 0x90
; XOPAVX2-NEXT: .LBB8_1: # %loop
; XOPAVX2-NEXT: # =>This Inner Loop Header: Depth=1
; XOPAVX2-NEXT: vprotd %xmm0, 1024(%rdi,%rax), %xmm1
; XOPAVX2-NEXT: vmovdqu %xmm1, 1024(%rdi,%rax)
; XOPAVX2-NEXT: addq $16, %rax
; XOPAVX2-NEXT: jne .LBB8_1
; XOPAVX2-NEXT: # %bb.2: # %end
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: sink_splatvar:
; X86-SSE2: # %bb.0: # %entry
; X86-SSE2-NEXT: pushl %esi
; X86-SSE2-NEXT: .cfi_def_cfa_offset 8
; X86-SSE2-NEXT: .cfi_offset %esi, -8
; X86-SSE2-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-SSE2-NEXT: xorl %ecx, %ecx
; X86-SSE2-NEXT: movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE2-NEXT: xorl %edx, %edx
; X86-SSE2-NEXT: .p2align 4, 0x90
; X86-SSE2-NEXT: .LBB8_1: # %loop
; X86-SSE2-NEXT: # =>This Inner Loop Header: Depth=1
; X86-SSE2-NEXT: movdqu (%eax,%ecx,4), %xmm1
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,2,3,3]
; X86-SSE2-NEXT: psllq %xmm0, %xmm2
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
; X86-SSE2-NEXT: psllq %xmm0, %xmm1
; X86-SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[1,3],xmm2[1,3]
; X86-SSE2-NEXT: movups %xmm1, (%eax,%ecx,4)
; X86-SSE2-NEXT: addl $4, %ecx
; X86-SSE2-NEXT: adcl $0, %edx
; X86-SSE2-NEXT: movl %ecx, %esi
; X86-SSE2-NEXT: xorl $256, %esi # imm = 0x100
; X86-SSE2-NEXT: orl %edx, %esi
; X86-SSE2-NEXT: jne .LBB8_1
; X86-SSE2-NEXT: # %bb.2: # %end
; X86-SSE2-NEXT: popl %esi
; X86-SSE2-NEXT: .cfi_def_cfa_offset 4
; X86-SSE2-NEXT: retl
entry:
%ins = insertelement <4 x i32> undef, i32 %shift_amt, i32 0
%splat = shufflevector <4 x i32> %ins, <4 x i32> undef, <4 x i32> zeroinitializer
br label %loop
loop:
%index = phi i64 [ 0, %entry ], [ %inc, %loop ]
%addr = getelementptr inbounds i32, ptr %p, i64 %index
%x = load <4 x i32>, ptr %addr, align 4
%fsh = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %x, <4 x i32> %x, <4 x i32> %splat)
store <4 x i32> %fsh, ptr %addr, align 4
%inc = add i64 %index, 4
%iv = icmp eq i64 %inc, 256
br i1 %iv, label %end, label %loop
end:
ret void
}
;
; Constant Shifts
;
define <2 x i64> @constant_funnnel_v2i64(<2 x i64> %x, <2 x i64> %y) nounwind {
; SSE2-LABEL: constant_funnnel_v2i64:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm1, %xmm2
; SSE2-NEXT: psrlq $60, %xmm2
; SSE2-NEXT: psrlq $50, %xmm1
; SSE2-NEXT: shufpd {{.*#+}} xmm2 = xmm2[0],xmm1[1]
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: psllq $4, %xmm1
; SSE2-NEXT: psllq $14, %xmm0
; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE2-NEXT: orpd %xmm2, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: constant_funnnel_v2i64:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm1, %xmm2
; SSE41-NEXT: psrlq $50, %xmm2
; SSE41-NEXT: psrlq $60, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm1[0,1,2,3],xmm2[4,5,6,7]
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: psllq $14, %xmm1
; SSE41-NEXT: psllq $4, %xmm0
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT: por %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: constant_funnnel_v2i64:
; AVX1: # %bb.0:
; AVX1-NEXT: vpsrlq $50, %xmm1, %xmm2
; AVX1-NEXT: vpsrlq $60, %xmm1, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm2[4,5,6,7]
; AVX1-NEXT: vpsllq $14, %xmm0, %xmm2
; AVX1-NEXT: vpsllq $4, %xmm0, %xmm0
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
; AVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: constant_funnnel_v2i64:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX2-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; AVX512F-LABEL: constant_funnnel_v2i64:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512F-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: constant_funnnel_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: constant_funnnel_v2i64:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512BW-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: constant_funnnel_v2i64:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpmovsxbq {{.*#+}} xmm2 = [4,14]
; AVX512VBMI2-NEXT: vpshldvq %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: constant_funnnel_v2i64:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: constant_funnnel_v2i64:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: constant_funnnel_v2i64:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshlq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX1-NEXT: vpshlq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: constant_funnnel_v2i64:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsrlvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX2-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: constant_funnnel_v2i64:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm2
; X86-SSE2-NEXT: psrlq $60, %xmm2
; X86-SSE2-NEXT: psrlq $50, %xmm1
; X86-SSE2-NEXT: shufpd {{.*#+}} xmm2 = xmm2[0],xmm1[1]
; X86-SSE2-NEXT: movdqa %xmm0, %xmm1
; X86-SSE2-NEXT: psllq $4, %xmm1
; X86-SSE2-NEXT: psllq $14, %xmm0
; X86-SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X86-SSE2-NEXT: orpd %xmm2, %xmm0
; X86-SSE2-NEXT: retl
%res = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> <i64 4, i64 14>)
ret <2 x i64> %res
}
define <4 x i32> @constant_funnnel_v4i32(<4 x i32> %x, <4 x i32> %y) nounwind {
; SSE2-LABEL: constant_funnnel_v4i32:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm1, %xmm2
; SSE2-NEXT: psrld $25, %xmm2
; SSE2-NEXT: movdqa %xmm1, %xmm3
; SSE2-NEXT: psrld $26, %xmm3
; SSE2-NEXT: punpckhqdq {{.*#+}} xmm3 = xmm3[1],xmm2[1]
; SSE2-NEXT: movdqa %xmm1, %xmm2
; SSE2-NEXT: psrld $27, %xmm2
; SSE2-NEXT: psrld $28, %xmm1
; SSE2-NEXT: punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,3],xmm3[0,3]
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; SSE2-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; SSE2-NEXT: por %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: constant_funnnel_v4i32:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm1, %xmm2
; SSE41-NEXT: psrld $25, %xmm2
; SSE41-NEXT: movdqa %xmm1, %xmm3
; SSE41-NEXT: psrld $27, %xmm3
; SSE41-NEXT: pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm2[4,5,6,7]
; SSE41-NEXT: movdqa %xmm1, %xmm2
; SSE41-NEXT: psrld $26, %xmm2
; SSE41-NEXT: psrld $28, %xmm1
; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm1[0,1,2,3],xmm2[4,5,6,7]
; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm3[2,3],xmm2[4,5],xmm3[6,7]
; SSE41-NEXT: pmulld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE41-NEXT: por %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: constant_funnnel_v4i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vpsrld $25, %xmm1, %xmm2
; AVX1-NEXT: vpsrld $27, %xmm1, %xmm3
; AVX1-NEXT: vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
; AVX1-NEXT: vpsrld $26, %xmm1, %xmm3
; AVX1-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm3[4,5,6,7]
; AVX1-NEXT: vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
; AVX1-NEXT: vpmulld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: constant_funnnel_v4i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX2-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; AVX512F-LABEL: constant_funnnel_v4i32:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512F-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: constant_funnnel_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: constant_funnnel_v4i32:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512BW-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: constant_funnnel_v4i32:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpmovsxbd {{.*#+}} xmm2 = [4,5,6,7]
; AVX512VBMI2-NEXT: vpshldvd %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: constant_funnnel_v4i32:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: constant_funnnel_v4i32:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOPAVX1-LABEL: constant_funnnel_v4i32:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX1-NEXT: vpshld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX1-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: constant_funnnel_v4i32:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsrlvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX2-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX2-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; X86-SSE2-LABEL: constant_funnnel_v4i32:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm2
; X86-SSE2-NEXT: psrld $25, %xmm2
; X86-SSE2-NEXT: movdqa %xmm1, %xmm3
; X86-SSE2-NEXT: psrld $26, %xmm3
; X86-SSE2-NEXT: punpckhqdq {{.*#+}} xmm3 = xmm3[1],xmm2[1]
; X86-SSE2-NEXT: movdqa %xmm1, %xmm2
; X86-SSE2-NEXT: psrld $27, %xmm2
; X86-SSE2-NEXT: psrld $28, %xmm1
; X86-SSE2-NEXT: punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
; X86-SSE2-NEXT: shufps {{.*#+}} xmm1 = xmm1[0,3],xmm3[0,3]
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; X86-SSE2-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X86-SSE2-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
; X86-SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
ret <4 x i32> %res
}
define <8 x i16> @constant_funnnel_v8i16(<8 x i16> %x, <8 x i16> %y) nounwind {
; SSE-LABEL: constant_funnnel_v8i16:
; SSE: # %bb.0:
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 # [1,2,4,8,16,32,64,128]
; SSE-NEXT: psrlw $1, %xmm1
; SSE-NEXT: pmulhuw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 # [2,4,8,16,32,64,128,256]
; SSE-NEXT: por %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: constant_funnnel_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX-NEXT: vpmulhuw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 # [2,4,8,16,32,64,128,256]
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: constant_funnnel_v8i16:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX512F-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512F-NEXT: vpmulhuw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 # [2,4,8,16,32,64,128,256]
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: constant_funnnel_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX512VL-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VL-NEXT: vpmulhuw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 # [2,4,8,16,32,64,128,256]
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: constant_funnnel_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT: vpmovsxbw {{.*#+}} xmm2 = [0,1,2,3,4,5,6,7]
; AVX512BW-NEXT: vpsllvw %zmm2, %zmm0, %zmm0
; AVX512BW-NEXT: vpmovsxbw {{.*#+}} xmm2 = [15,14,13,12,11,10,9,8]
; AVX512BW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512BW-NEXT: vpsrlvw %zmm2, %zmm1, %zmm1
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: constant_funnnel_v8i16:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpmovsxbw {{.*#+}} xmm2 = [0,1,2,3,4,5,6,7]
; AVX512VBMI2-NEXT: vpshldvw %zmm2, %zmm1, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: constant_funnnel_v8i16:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT: vpsrlw $1, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsrlvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: constant_funnnel_v8i16:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: constant_funnnel_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpshlw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOP-NEXT: vpsrlw $1, %xmm1, %xmm1
; XOP-NEXT: vpshlw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: constant_funnnel_v8i16:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0 # [1,2,4,8,16,32,64,128]
; X86-SSE2-NEXT: psrlw $1, %xmm1
; X86-SSE2-NEXT: pmulhuw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 # [2,4,8,16,32,64,128,256]
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
ret <8 x i16> %res
}
define <16 x i8> @constant_funnnel_v16i8(<16 x i8> %x, <16 x i8> %y) nounwind {
; SSE-LABEL: constant_funnnel_v16i8:
; SSE: # %bb.0:
; SSE-NEXT: movdqa %xmm1, %xmm2
; SSE-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2 # [1,128,64,32,16,8,4,2]
; SSE-NEXT: psrlw $8, %xmm2
; SSE-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1 # [1,2,4,8,16,32,64,128]
; SSE-NEXT: psrlw $8, %xmm1
; SSE-NEXT: packuswb %xmm2, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: constant_funnnel_v16i8:
; AVX: # %bb.0:
; AVX-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2 # [1,128,64,32,16,8,4,2]
; AVX-NEXT: vpsrlw $8, %xmm2, %xmm2
; AVX-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX-NEXT: vpackuswb %xmm2, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: constant_funnnel_v16i8:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512F-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2 # [1,128,64,32,16,8,4,2]
; AVX512F-NEXT: vpsrlw $8, %xmm2, %xmm2
; AVX512F-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512F-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX512F-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512F-NEXT: vpackuswb %xmm2, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: constant_funnnel_v16i8:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpunpckhbw {{.*#+}} xmm2 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
; AVX512VL-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2 # [1,128,64,32,16,8,4,2]
; AVX512VL-NEXT: vpsrlw $8, %xmm2, %xmm2
; AVX512VL-NEXT: vpunpcklbw {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; AVX512VL-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 # [1,2,4,8,16,32,64,128]
; AVX512VL-NEXT: vpsrlw $8, %xmm0, %xmm0
; AVX512VL-NEXT: vpackuswb %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: constant_funnnel_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpmovsxbw {{.*#+}} ymm2 = [0,1,2,3,4,5,6,7,0,7,6,5,4,3,2,1]
; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512BW-NEXT: vpsllw $8, %ymm0, %ymm0
; AVX512BW-NEXT: vpor %ymm1, %ymm0, %ymm0
; AVX512BW-NEXT: vpsllvw %zmm2, %zmm0, %zmm0
; AVX512BW-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: constant_funnnel_v16i8:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vmovdqa {{.*#+}} ymm2 = [0,64,1,65,2,66,3,67,4,68,5,69,6,70,7,71,8,72,9,73,10,74,11,75,12,76,13,77,14,78,15,79]
; AVX512VBMI2-NEXT: vpermt2b %zmm0, %zmm2, %zmm1
; AVX512VBMI2-NEXT: vpmovsxbw {{.*#+}} ymm0 = [0,1,2,3,4,5,6,7,0,7,6,5,4,3,2,1]
; AVX512VBMI2-NEXT: vpsllvw %zmm0, %zmm1, %zmm0
; AVX512VBMI2-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VBMI2-NEXT: vpmovwb %zmm0, %ymm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: constant_funnnel_v16i8:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
; AVX512VLBW-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; AVX512VLBW-NEXT: vpsllw $8, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpor %ymm1, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX512VLBW-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VLBW-NEXT: vpmovwb %ymm0, %xmm0
; AVX512VLBW-NEXT: vzeroupper
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: constant_funnnel_v16i8:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $ymm1
; AVX512VLVBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0
; AVX512VLVBMI2-NEXT: vmovdqa {{.*#+}} ymm2 = [0,32,1,33,2,34,3,35,4,36,5,37,6,38,7,39,8,40,9,41,10,42,11,43,12,44,13,45,14,46,15,47]
; AVX512VLVBMI2-NEXT: vpermi2b %ymm0, %ymm1, %ymm2
; AVX512VLVBMI2-NEXT: vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm0
; AVX512VLVBMI2-NEXT: vpsrlw $8, %ymm0, %ymm0
; AVX512VLVBMI2-NEXT: vpmovwb %ymm0, %xmm0
; AVX512VLVBMI2-NEXT: vzeroupper
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: constant_funnnel_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; XOP-NEXT: vpshlb %xmm2, %xmm1, %xmm1
; XOP-NEXT: vpshlb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOP-NEXT: vpshlb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: constant_funnnel_v16i8:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: movdqa %xmm1, %xmm2
; X86-SSE2-NEXT: punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
; X86-SSE2-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2 # [1,128,64,32,16,8,4,2]
; X86-SSE2-NEXT: psrlw $8, %xmm2
; X86-SSE2-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
; X86-SSE2-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1 # [1,2,4,8,16,32,64,128]
; X86-SSE2-NEXT: psrlw $8, %xmm1
; X86-SSE2-NEXT: packuswb %xmm2, %xmm1
; X86-SSE2-NEXT: movdqa %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>)
ret <16 x i8> %res
}
;
; Uniform Constant Shifts
;
define <2 x i64> @splatconstant_funnnel_v2i64(<2 x i64> %x, <2 x i64> %y) nounwind {
; SSE-LABEL: splatconstant_funnnel_v2i64:
; SSE: # %bb.0:
; SSE-NEXT: psrlq $50, %xmm1
; SSE-NEXT: psllq $14, %xmm0
; SSE-NEXT: por %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_funnnel_v2i64:
; AVX: # %bb.0:
; AVX-NEXT: vpsrlq $50, %xmm1, %xmm1
; AVX-NEXT: vpsllq $14, %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatconstant_funnnel_v2i64:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsrlq $50, %xmm1, %xmm1
; AVX512F-NEXT: vpsllq $14, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_funnnel_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsrlq $50, %xmm1, %xmm1
; AVX512VL-NEXT: vpsllq $14, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatconstant_funnnel_v2i64:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsrlq $50, %xmm1, %xmm1
; AVX512BW-NEXT: vpsllq $14, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatconstant_funnnel_v2i64:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldq $14, %zmm1, %zmm0, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatconstant_funnnel_v2i64:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsrlq $50, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsllq $14, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatconstant_funnnel_v2i64:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldq $14, %xmm1, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatconstant_funnnel_v2i64:
; XOP: # %bb.0:
; XOP-NEXT: vpsrlq $50, %xmm1, %xmm1
; XOP-NEXT: vpsllq $14, %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatconstant_funnnel_v2i64:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: psrlq $50, %xmm1
; X86-SSE2-NEXT: psllq $14, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %x, <2 x i64> %y, <2 x i64> <i64 14, i64 14>)
ret <2 x i64> %res
}
define <4 x i32> @splatconstant_funnnel_v4i32(<4 x i32> %x, <4 x i32> %y) nounwind {
; SSE-LABEL: splatconstant_funnnel_v4i32:
; SSE: # %bb.0:
; SSE-NEXT: psrld $28, %xmm1
; SSE-NEXT: pslld $4, %xmm0
; SSE-NEXT: por %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_funnnel_v4i32:
; AVX: # %bb.0:
; AVX-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX-NEXT: vpslld $4, %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatconstant_funnnel_v4i32:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX512F-NEXT: vpslld $4, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_funnnel_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX512VL-NEXT: vpslld $4, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatconstant_funnnel_v4i32:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX512BW-NEXT: vpslld $4, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatconstant_funnnel_v4i32:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldd $4, %zmm1, %zmm0, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatconstant_funnnel_v4i32:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsrld $28, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpslld $4, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatconstant_funnnel_v4i32:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldd $4, %xmm1, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatconstant_funnnel_v4i32:
; XOP: # %bb.0:
; XOP-NEXT: vpsrld $28, %xmm1, %xmm1
; XOP-NEXT: vpslld $4, %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatconstant_funnnel_v4i32:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: psrld $28, %xmm1
; X86-SSE2-NEXT: pslld $4, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %x, <4 x i32> %y, <4 x i32> <i32 4, i32 4, i32 4, i32 4>)
ret <4 x i32> %res
}
define <8 x i16> @splatconstant_funnnel_v8i16(<8 x i16> %x, <8 x i16> %y) nounwind {
; SSE-LABEL: splatconstant_funnnel_v8i16:
; SSE: # %bb.0:
; SSE-NEXT: psrlw $9, %xmm1
; SSE-NEXT: psllw $7, %xmm0
; SSE-NEXT: por %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_funnnel_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpsrlw $9, %xmm1, %xmm1
; AVX-NEXT: vpsllw $7, %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatconstant_funnnel_v8i16:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsrlw $9, %xmm1, %xmm1
; AVX512F-NEXT: vpsllw $7, %xmm0, %xmm0
; AVX512F-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_funnnel_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsrlw $9, %xmm1, %xmm1
; AVX512VL-NEXT: vpsllw $7, %xmm0, %xmm0
; AVX512VL-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatconstant_funnnel_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsrlw $9, %xmm1, %xmm1
; AVX512BW-NEXT: vpsllw $7, %xmm0, %xmm0
; AVX512BW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatconstant_funnnel_v8i16:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512VBMI2-NEXT: vpshldw $7, %zmm1, %zmm0, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatconstant_funnnel_v8i16:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsrlw $9, %xmm1, %xmm1
; AVX512VLBW-NEXT: vpsllw $7, %xmm0, %xmm0
; AVX512VLBW-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatconstant_funnnel_v8i16:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpshldw $7, %xmm1, %xmm0, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatconstant_funnnel_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpsrlw $9, %xmm1, %xmm1
; XOP-NEXT: vpsllw $7, %xmm0, %xmm0
; XOP-NEXT: vpor %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatconstant_funnnel_v8i16:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: psrlw $9, %xmm1
; X86-SSE2-NEXT: psllw $7, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %x, <8 x i16> %y, <8 x i16> <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>)
ret <8 x i16> %res
}
define <16 x i8> @splatconstant_funnnel_v16i8(<16 x i8> %x, <16 x i8> %y) nounwind {
; SSE-LABEL: splatconstant_funnnel_v16i8:
; SSE: # %bb.0:
; SSE-NEXT: psrlw $4, %xmm1
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE-NEXT: psllw $4, %xmm0
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: por %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_funnnel_v16i8:
; AVX: # %bb.0:
; AVX-NEXT: vpsrlw $4, %xmm1, %xmm1
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX-NEXT: vpsllw $4, %xmm0, %xmm0
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; AVX512F-LABEL: splatconstant_funnnel_v16i8:
; AVX512F: # %bb.0:
; AVX512F-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512F-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512F-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm2, %zmm0
; AVX512F-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT: vzeroupper
; AVX512F-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_funnnel_v16i8:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512VL-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512VL-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm0
; AVX512VL-NEXT: retq
;
; AVX512BW-LABEL: splatconstant_funnnel_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512BW-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512BW-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm2, %zmm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512VBMI2-LABEL: splatconstant_funnnel_v16i8:
; AVX512VBMI2: # %bb.0:
; AVX512VBMI2-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512VBMI2-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512VBMI2-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm2, %zmm0
; AVX512VBMI2-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512VBMI2-NEXT: vzeroupper
; AVX512VBMI2-NEXT: retq
;
; AVX512VLBW-LABEL: splatconstant_funnnel_v16i8:
; AVX512VLBW: # %bb.0:
; AVX512VLBW-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512VLBW-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512VLBW-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm0
; AVX512VLBW-NEXT: retq
;
; AVX512VLVBMI2-LABEL: splatconstant_funnnel_v16i8:
; AVX512VLVBMI2: # %bb.0:
; AVX512VLVBMI2-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX512VLVBMI2-NEXT: vpsrlw $4, %xmm1, %xmm0
; AVX512VLVBMI2-NEXT: vpternlogd $216, {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm2, %xmm0
; AVX512VLVBMI2-NEXT: retq
;
; XOP-LABEL: splatconstant_funnnel_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpsrlw $4, %xmm1, %xmm1
; XOP-NEXT: vpsllw $4, %xmm0, %xmm0
; XOP-NEXT: vpcmov {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; X86-SSE2-LABEL: splatconstant_funnnel_v16i8:
; X86-SSE2: # %bb.0:
; X86-SSE2-NEXT: psrlw $4, %xmm1
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE2-NEXT: psllw $4, %xmm0
; X86-SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE2-NEXT: por %xmm1, %xmm0
; X86-SSE2-NEXT: retl
%res = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %x, <16 x i8> %y, <16 x i8> <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>)
ret <16 x i8> %res
}
Codebase Browser
llvm