; 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=+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
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq | FileCheck %s --check-prefixes=AVX512,AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw | FileCheck %s --check-prefixes=AVX512,AVX512BW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq,+avx512vl | FileCheck %s --check-prefixes=AVX512VL,AVX512DQVL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512VL,AVX512BWVL
;
; Just one 32-bit run to make sure we do reasonable things for i64 shifts.
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X86-SSE
;
; Variable Shifts
;
define <2 x i64> @var_shift_v2i64(<2 x i64> %a, <2 x i64> %b) nounwind {
; SSE2-LABEL: var_shift_v2i64:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm0, %xmm2
; SSE2-NEXT: psllq %xmm1, %xmm2
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
; SSE2-NEXT: psllq %xmm1, %xmm0
; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_shift_v2i64:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm0, %xmm2
; SSE41-NEXT: psllq %xmm1, %xmm2
; SSE41-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
; SSE41-NEXT: psllq %xmm1, %xmm0
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm2[0,1,2,3],xmm0[4,5,6,7]
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_shift_v2i64:
; AVX1: # %bb.0:
; AVX1-NEXT: vpsllq %xmm1, %xmm0, %xmm2
; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
; AVX1-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm2[0,1,2,3],xmm0[4,5,6,7]
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_shift_v2i64:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsllvq %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: var_shift_v2i64:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshlq %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_shift_v2i64:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsllvq %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: var_shift_v2i64:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvq %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: var_shift_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllvq %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: var_shift_v2i64:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: movdqa %xmm0, %xmm2
; X86-SSE-NEXT: psllq %xmm1, %xmm2
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
; X86-SSE-NEXT: psllq %xmm1, %xmm0
; X86-SSE-NEXT: movsd {{.*#+}} xmm0 = xmm2[0],xmm0[1]
; X86-SSE-NEXT: retl
%shift = shl <2 x i64> %a, %b
ret <2 x i64> %shift
}
define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) nounwind {
; SSE2-LABEL: var_shift_v4i32:
; SSE2: # %bb.0:
; SSE2-NEXT: pslld $23, %xmm1
; SSE2-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE2-NEXT: cvttps2dq %xmm1, %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: retq
;
; SSE41-LABEL: var_shift_v4i32:
; SSE41: # %bb.0:
; SSE41-NEXT: pslld $23, %xmm1
; SSE41-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE41-NEXT: cvttps2dq %xmm1, %xmm1
; SSE41-NEXT: pmulld %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_shift_v4i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vpslld $23, %xmm1, %xmm1
; AVX1-NEXT: vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX1-NEXT: vcvttps2dq %xmm1, %xmm1
; AVX1-NEXT: vpmulld %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_shift_v4i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsllvd %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: var_shift_v4i32:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshld %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: var_shift_v4i32:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsllvd %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: var_shift_v4i32:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvd %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: var_shift_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllvd %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: var_shift_v4i32:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pslld $23, %xmm1
; X86-SSE-NEXT: paddd {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: cvttps2dq %xmm1, %xmm1
; X86-SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; X86-SSE-NEXT: pmuludq %xmm1, %xmm0
; X86-SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; X86-SSE-NEXT: pmuludq %xmm2, %xmm1
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X86-SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE-NEXT: retl
%shift = shl <4 x i32> %a, %b
ret <4 x i32> %shift
}
define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; SSE2-LABEL: var_shift_v8i16:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm1, %xmm2
; SSE2-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4,4,5,5,6,6,7,7]
; SSE2-NEXT: pslld $23, %xmm2
; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; SSE2-NEXT: paddd %xmm3, %xmm2
; SSE2-NEXT: cvttps2dq %xmm2, %xmm2
; SSE2-NEXT: pslld $16, %xmm2
; SSE2-NEXT: psrad $16, %xmm2
; SSE2-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3]
; SSE2-NEXT: pslld $23, %xmm1
; SSE2-NEXT: paddd %xmm3, %xmm1
; SSE2-NEXT: cvttps2dq %xmm1, %xmm1
; SSE2-NEXT: pslld $16, %xmm1
; SSE2-NEXT: psrad $16, %xmm1
; SSE2-NEXT: packssdw %xmm2, %xmm1
; SSE2-NEXT: pmullw %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_shift_v8i16:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovzxwd {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
; SSE41-NEXT: punpckhwd {{.*#+}} xmm1 = xmm1[4,4,5,5,6,6,7,7]
; SSE41-NEXT: pslld $23, %xmm1
; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; SSE41-NEXT: paddd %xmm3, %xmm1
; SSE41-NEXT: cvttps2dq %xmm1, %xmm1
; SSE41-NEXT: pslld $23, %xmm2
; SSE41-NEXT: paddd %xmm3, %xmm2
; SSE41-NEXT: cvttps2dq %xmm2, %xmm2
; SSE41-NEXT: packusdw %xmm1, %xmm2
; SSE41-NEXT: pmullw %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: var_shift_v8i16:
; AVX1: # %bb.0:
; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm1[4,4,5,5,6,6,7,7]
; AVX1-NEXT: vpslld $23, %xmm2, %xmm2
; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; AVX1-NEXT: vpaddd %xmm3, %xmm2, %xmm2
; AVX1-NEXT: vcvttps2dq %xmm2, %xmm2
; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero
; AVX1-NEXT: vpslld $23, %xmm1, %xmm1
; AVX1-NEXT: vpaddd %xmm3, %xmm1, %xmm1
; AVX1-NEXT: vcvttps2dq %xmm1, %xmm1
; AVX1-NEXT: vpackusdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpmullw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: var_shift_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: vpsllvd %ymm1, %ymm0, %ymm0
; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,u,u,u,u,u,u,u,u,16,17,20,21,24,25,28,29,u,u,u,u,u,u,u,u]
; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; XOP-LABEL: var_shift_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpshlw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512DQ-LABEL: var_shift_v8i16:
; AVX512DQ: # %bb.0:
; AVX512DQ-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
; AVX512DQ-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
; AVX512DQ-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: vpmovdw %zmm0, %ymm0
; AVX512DQ-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512DQ-NEXT: vzeroupper
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: var_shift_v8i16:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-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
; AVX512DQVL-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
; AVX512DQVL-NEXT: vpsllvd %ymm1, %ymm0, %ymm0
; AVX512DQVL-NEXT: vpmovdw %ymm0, %xmm0
; AVX512DQVL-NEXT: vzeroupper
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: var_shift_v8i16:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: vpsllvw %xmm1, %xmm0, %xmm0
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: var_shift_v8i16:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: movdqa %xmm1, %xmm2
; X86-SSE-NEXT: punpckhwd {{.*#+}} xmm2 = xmm2[4,4,5,5,6,6,7,7]
; X86-SSE-NEXT: pslld $23, %xmm2
; X86-SSE-NEXT: movdqa {{.*#+}} xmm3 = [1065353216,1065353216,1065353216,1065353216]
; X86-SSE-NEXT: paddd %xmm3, %xmm2
; X86-SSE-NEXT: cvttps2dq %xmm2, %xmm2
; X86-SSE-NEXT: pslld $16, %xmm2
; X86-SSE-NEXT: psrad $16, %xmm2
; X86-SSE-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3]
; X86-SSE-NEXT: pslld $23, %xmm1
; X86-SSE-NEXT: paddd %xmm3, %xmm1
; X86-SSE-NEXT: cvttps2dq %xmm1, %xmm1
; X86-SSE-NEXT: pslld $16, %xmm1
; X86-SSE-NEXT: psrad $16, %xmm1
; X86-SSE-NEXT: packssdw %xmm2, %xmm1
; X86-SSE-NEXT: pmullw %xmm1, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <8 x i16> %a, %b
ret <8 x i16> %shift
}
define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind {
; SSE2-LABEL: var_shift_v16i8:
; SSE2: # %bb.0:
; SSE2-NEXT: psllw $5, %xmm1
; SSE2-NEXT: pxor %xmm2, %xmm2
; SSE2-NEXT: pxor %xmm3, %xmm3
; SSE2-NEXT: pcmpgtb %xmm1, %xmm3
; SSE2-NEXT: movdqa %xmm3, %xmm4
; SSE2-NEXT: pandn %xmm0, %xmm4
; SSE2-NEXT: psllw $4, %xmm0
; SSE2-NEXT: pand %xmm3, %xmm0
; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT: por %xmm4, %xmm0
; SSE2-NEXT: paddb %xmm1, %xmm1
; SSE2-NEXT: pxor %xmm3, %xmm3
; SSE2-NEXT: pcmpgtb %xmm1, %xmm3
; SSE2-NEXT: movdqa %xmm3, %xmm4
; SSE2-NEXT: pandn %xmm0, %xmm4
; SSE2-NEXT: psllw $2, %xmm0
; SSE2-NEXT: pand %xmm3, %xmm0
; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT: por %xmm4, %xmm0
; SSE2-NEXT: paddb %xmm1, %xmm1
; SSE2-NEXT: pcmpgtb %xmm1, %xmm2
; SSE2-NEXT: movdqa %xmm2, %xmm1
; SSE2-NEXT: pandn %xmm0, %xmm1
; SSE2-NEXT: paddb %xmm0, %xmm0
; SSE2-NEXT: pand %xmm2, %xmm0
; SSE2-NEXT: por %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: var_shift_v16i8:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm0, %xmm2
; SSE41-NEXT: psllw $5, %xmm1
; SSE41-NEXT: movdqa %xmm0, %xmm3
; SSE41-NEXT: psllw $4, %xmm3
; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3
; SSE41-NEXT: movdqa %xmm1, %xmm0
; SSE41-NEXT: pblendvb %xmm0, %xmm3, %xmm2
; SSE41-NEXT: movdqa %xmm2, %xmm3
; SSE41-NEXT: psllw $2, %xmm3
; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3
; SSE41-NEXT: paddb %xmm1, %xmm1
; SSE41-NEXT: movdqa %xmm1, %xmm0
; SSE41-NEXT: pblendvb %xmm0, %xmm3, %xmm2
; SSE41-NEXT: movdqa %xmm2, %xmm3
; SSE41-NEXT: paddb %xmm2, %xmm3
; SSE41-NEXT: paddb %xmm1, %xmm1
; SSE41-NEXT: movdqa %xmm1, %xmm0
; SSE41-NEXT: pblendvb %xmm0, %xmm3, %xmm2
; SSE41-NEXT: movdqa %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: var_shift_v16i8:
; AVX: # %bb.0:
; AVX-NEXT: vpsllw $5, %xmm1, %xmm1
; AVX-NEXT: vpsllw $4, %xmm0, %xmm2
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpsllw $2, %xmm0, %xmm2
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2
; AVX-NEXT: vpaddb %xmm1, %xmm1, %xmm1
; AVX-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT: vpaddb %xmm0, %xmm0, %xmm2
; AVX-NEXT: vpaddb %xmm1, %xmm1, %xmm1
; AVX-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: var_shift_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpshlb %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512DQ-LABEL: var_shift_v16i8:
; AVX512DQ: # %bb.0:
; AVX512DQ-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
; AVX512DQ-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
; AVX512DQ-NEXT: vpsllvd %zmm1, %zmm0, %zmm0
; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQ-NEXT: vzeroupper
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v16i8:
; AVX512BW: # %bb.0:
; 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: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: var_shift_v16i8:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-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
; AVX512DQVL-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
; AVX512DQVL-NEXT: vpsllvd %zmm1, %zmm0, %zmm0
; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQVL-NEXT: vzeroupper
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: var_shift_v16i8:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-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
; AVX512BWVL-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
; AVX512BWVL-NEXT: vpsllvw %ymm1, %ymm0, %ymm0
; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0
; AVX512BWVL-NEXT: vzeroupper
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: var_shift_v16i8:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: psllw $5, %xmm1
; X86-SSE-NEXT: pxor %xmm2, %xmm2
; X86-SSE-NEXT: pxor %xmm3, %xmm3
; X86-SSE-NEXT: pcmpgtb %xmm1, %xmm3
; X86-SSE-NEXT: movdqa %xmm3, %xmm4
; X86-SSE-NEXT: pandn %xmm0, %xmm4
; X86-SSE-NEXT: psllw $4, %xmm0
; X86-SSE-NEXT: pand %xmm3, %xmm0
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: por %xmm4, %xmm0
; X86-SSE-NEXT: paddb %xmm1, %xmm1
; X86-SSE-NEXT: pxor %xmm3, %xmm3
; X86-SSE-NEXT: pcmpgtb %xmm1, %xmm3
; X86-SSE-NEXT: movdqa %xmm3, %xmm4
; X86-SSE-NEXT: pandn %xmm0, %xmm4
; X86-SSE-NEXT: psllw $2, %xmm0
; X86-SSE-NEXT: pand %xmm3, %xmm0
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: por %xmm4, %xmm0
; X86-SSE-NEXT: paddb %xmm1, %xmm1
; X86-SSE-NEXT: pcmpgtb %xmm1, %xmm2
; X86-SSE-NEXT: movdqa %xmm2, %xmm1
; X86-SSE-NEXT: pandn %xmm0, %xmm1
; X86-SSE-NEXT: paddb %xmm0, %xmm0
; X86-SSE-NEXT: pand %xmm2, %xmm0
; X86-SSE-NEXT: por %xmm1, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <16 x i8> %a, %b
ret <16 x i8> %shift
}
;
; Uniform Variable Shifts
;
define <2 x i64> @splatvar_shift_v2i64(<2 x i64> %a, <2 x i64> %b) nounwind {
; SSE-LABEL: splatvar_shift_v2i64:
; SSE: # %bb.0:
; SSE-NEXT: psllq %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_shift_v2i64:
; AVX: # %bb.0:
; AVX-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_shift_v2i64:
; XOP: # %bb.0:
; XOP-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_shift_v2i64:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_shift_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_shift_v2i64:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: psllq %xmm1, %xmm0
; X86-SSE-NEXT: retl
%splat = shufflevector <2 x i64> %b, <2 x i64> undef, <2 x i32> zeroinitializer
%shift = shl <2 x i64> %a, %splat
ret <2 x i64> %shift
}
define <4 x i32> @splatvar_shift_v4i32(<4 x i32> %a, <4 x i32> %b) nounwind {
; SSE2-LABEL: splatvar_shift_v4i32:
; SSE2: # %bb.0:
; SSE2-NEXT: xorps %xmm2, %xmm2
; SSE2-NEXT: movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
; SSE2-NEXT: pslld %xmm2, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_shift_v4i32:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; SSE41-NEXT: pslld %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: splatvar_shift_v4i32:
; AVX: # %bb.0:
; AVX-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; AVX-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_shift_v4i32:
; XOP: # %bb.0:
; XOP-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; XOP-NEXT: vpslld %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_shift_v4i32:
; AVX512: # %bb.0:
; AVX512-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; AVX512-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_shift_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; AVX512VL-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_shift_v4i32:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: xorps %xmm2, %xmm2
; X86-SSE-NEXT: movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
; X86-SSE-NEXT: pslld %xmm2, %xmm0
; X86-SSE-NEXT: retl
%splat = shufflevector <4 x i32> %b, <4 x i32> undef, <4 x i32> zeroinitializer
%shift = shl <4 x i32> %a, %splat
ret <4 x i32> %shift
}
define <8 x i16> @splatvar_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; SSE2-LABEL: splatvar_shift_v8i16:
; SSE2: # %bb.0:
; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE2-NEXT: psllw %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_shift_v8i16:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; SSE41-NEXT: psllw %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX-LABEL: splatvar_shift_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_shift_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; XOP-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_shift_v8i16:
; AVX512: # %bb.0:
; AVX512-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX512-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_shift_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX512VL-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_shift_v8i16:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; X86-SSE-NEXT: psrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; X86-SSE-NEXT: psllw %xmm1, %xmm0
; X86-SSE-NEXT: retl
%splat = shufflevector <8 x i16> %b, <8 x i16> undef, <8 x i32> zeroinitializer
%shift = shl <8 x i16> %a, %splat
ret <8 x i16> %shift
}
define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind {
; SSE2-LABEL: splatvar_shift_v16i8:
; SSE2: # %bb.0:
; SSE2-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0]
; SSE2-NEXT: psrldq {{.*#+}} xmm1 = xmm1[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE2-NEXT: psllw %xmm1, %xmm0
; SSE2-NEXT: pcmpeqd %xmm2, %xmm2
; SSE2-NEXT: psllw %xmm1, %xmm2
; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; SSE2-NEXT: pand %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_shift_v16i8:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; SSE41-NEXT: psllw %xmm1, %xmm0
; SSE41-NEXT: pcmpeqd %xmm2, %xmm2
; SSE41-NEXT: psllw %xmm1, %xmm2
; SSE41-NEXT: pxor %xmm1, %xmm1
; SSE41-NEXT: pshufb %xmm1, %xmm2
; SSE41-NEXT: pand %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: splatvar_shift_v16i8:
; AVX1: # %bb.0:
; AVX1-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX1-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpsllw %xmm1, %xmm2, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: splatvar_shift_v16i8:
; AVX2: # %bb.0:
; AVX2-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX2-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vpsllw %xmm1, %xmm2, %xmm1
; AVX2-NEXT: vpbroadcastb %xmm1, %xmm1
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: splatvar_shift_v16i8:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
; XOPAVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT: vpshlb %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: splatvar_shift_v16i8:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpbroadcastb %xmm1, %xmm1
; XOPAVX2-NEXT: vpshlb %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512DQ-LABEL: splatvar_shift_v16i8:
; AVX512DQ: # %bb.0:
; AVX512DQ-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512DQ-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
; AVX512DQ-NEXT: vpslld %xmm1, %zmm0, %zmm0
; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQ-NEXT: vzeroupper
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: splatvar_shift_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,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 %xmm1, %ymm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: splatvar_shift_v16i8:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512DQVL-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
; AVX512DQVL-NEXT: vpslld %xmm1, %zmm0, %zmm0
; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQVL-NEXT: vzeroupper
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: splatvar_shift_v16i8:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512BWVL-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
; AVX512BWVL-NEXT: vpsllw %xmm1, %ymm0, %ymm0
; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0
; AVX512BWVL-NEXT: vzeroupper
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_shift_v16i8:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0]
; X86-SSE-NEXT: psrldq {{.*#+}} xmm1 = xmm1[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; X86-SSE-NEXT: psllw %xmm1, %xmm0
; X86-SSE-NEXT: pcmpeqd %xmm2, %xmm2
; X86-SSE-NEXT: psllw %xmm1, %xmm2
; X86-SSE-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X86-SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,0,0,0,4,5,6,7]
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; X86-SSE-NEXT: pand %xmm1, %xmm0
; X86-SSE-NEXT: retl
%splat = shufflevector <16 x i8> %b, <16 x i8> undef, <16 x i32> zeroinitializer
%shift = shl <16 x i8> %a, %splat
ret <16 x i8> %shift
}
;
; Uniform Variable Modulo Shifts
;
define <2 x i64> @splatvar_modulo_shift_v2i64(<2 x i64> %a, <2 x i64> %b) nounwind {
; SSE-LABEL: splatvar_modulo_shift_v2i64:
; SSE: # %bb.0:
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE-NEXT: psllq %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_modulo_shift_v2i64:
; AVX: # %bb.0:
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_modulo_shift_v2i64:
; XOP: # %bb.0:
; XOP-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOP-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_modulo_shift_v2i64:
; AVX512: # %bb.0:
; AVX512-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_modulo_shift_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpandq {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to2}, %xmm1, %xmm1
; AVX512VL-NEXT: vpsllq %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_modulo_shift_v2i64:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: psllq %xmm1, %xmm0
; X86-SSE-NEXT: retl
%mod = and <2 x i64> %b, <i64 63, i64 63>
%splat = shufflevector <2 x i64> %mod, <2 x i64> undef, <2 x i32> zeroinitializer
%shift = shl <2 x i64> %a, %splat
ret <2 x i64> %shift
}
define <4 x i32> @splatvar_modulo_shift_v4i32(<4 x i32> %a, <4 x i32> %b) nounwind {
; SSE-LABEL: splatvar_modulo_shift_v4i32:
; SSE: # %bb.0:
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE-NEXT: pslld %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_modulo_shift_v4i32:
; AVX: # %bb.0:
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_modulo_shift_v4i32:
; XOP: # %bb.0:
; XOP-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOP-NEXT: vpslld %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_modulo_shift_v4i32:
; AVX512: # %bb.0:
; AVX512-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_modulo_shift_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT: vpslld %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_modulo_shift_v4i32:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: pslld %xmm1, %xmm0
; X86-SSE-NEXT: retl
%mod = and <4 x i32> %b, <i32 31, i32 31, i32 31, i32 31>
%splat = shufflevector <4 x i32> %mod, <4 x i32> undef, <4 x i32> zeroinitializer
%shift = shl <4 x i32> %a, %splat
ret <4 x i32> %shift
}
define <8 x i16> @splatvar_modulo_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; SSE-LABEL: splatvar_modulo_shift_v8i16:
; SSE: # %bb.0:
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE-NEXT: psllw %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatvar_modulo_shift_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatvar_modulo_shift_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOP-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatvar_modulo_shift_v8i16:
; AVX512: # %bb.0:
; AVX512-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatvar_modulo_shift_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_modulo_shift_v8i16:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: psllw %xmm1, %xmm0
; X86-SSE-NEXT: retl
%mod = and <8 x i16> %b, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
%splat = shufflevector <8 x i16> %mod, <8 x i16> undef, <8 x i32> zeroinitializer
%shift = shl <8 x i16> %a, %splat
ret <8 x i16> %shift
}
define <16 x i8> @splatvar_modulo_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind {
; SSE2-LABEL: splatvar_modulo_shift_v16i8:
; SSE2: # %bb.0:
; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE2-NEXT: psllw %xmm1, %xmm0
; SSE2-NEXT: pcmpeqd %xmm2, %xmm2
; SSE2-NEXT: psllw %xmm1, %xmm2
; SSE2-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; SSE2-NEXT: pand %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: splatvar_modulo_shift_v16i8:
; SSE41: # %bb.0:
; SSE41-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE41-NEXT: psllw %xmm1, %xmm0
; SSE41-NEXT: pcmpeqd %xmm2, %xmm2
; SSE41-NEXT: psllw %xmm1, %xmm2
; SSE41-NEXT: pxor %xmm1, %xmm1
; SSE41-NEXT: pshufb %xmm1, %xmm2
; SSE41-NEXT: pand %xmm2, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: splatvar_modulo_shift_v16i8:
; AVX1: # %bb.0:
; AVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX1-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpsllw %xmm1, %xmm2, %xmm1
; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: splatvar_modulo_shift_v16i8:
; AVX2: # %bb.0:
; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX2-NEXT: vpsllw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vpcmpeqd %xmm2, %xmm2, %xmm2
; AVX2-NEXT: vpsllw %xmm1, %xmm2, %xmm1
; AVX2-NEXT: vpbroadcastb %xmm1, %xmm1
; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: splatvar_modulo_shift_v16i8:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
; XOPAVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT: vpshlb %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: splatvar_modulo_shift_v16i8:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpbroadcastb %xmm1, %xmm1
; XOPAVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; XOPAVX2-NEXT: vpshlb %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512DQ-LABEL: splatvar_modulo_shift_v16i8:
; AVX512DQ: # %bb.0:
; AVX512DQ-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512DQ-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
; AVX512DQ-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512DQ-NEXT: vpslld %xmm1, %zmm0, %zmm0
; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQ-NEXT: vzeroupper
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: splatvar_modulo_shift_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; 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: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512BW-NEXT: vpsllw %xmm1, %ymm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: splatvar_modulo_shift_v16i8:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512DQVL-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
; AVX512DQVL-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512DQVL-NEXT: vpslld %xmm1, %zmm0, %zmm0
; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQVL-NEXT: vzeroupper
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: splatvar_modulo_shift_v16i8:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX512BWVL-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
; AVX512BWVL-NEXT: vpmovzxbq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,zero,zero,zero,zero,xmm1[1],zero,zero,zero,zero,zero,zero,zero
; AVX512BWVL-NEXT: vpsllw %xmm1, %ymm0, %ymm0
; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0
; AVX512BWVL-NEXT: vzeroupper
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: splatvar_modulo_shift_v16i8:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: psllw %xmm1, %xmm0
; X86-SSE-NEXT: pcmpeqd %xmm2, %xmm2
; X86-SSE-NEXT: psllw %xmm1, %xmm2
; X86-SSE-NEXT: punpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X86-SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,0,0,0,4,5,6,7]
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; X86-SSE-NEXT: pand %xmm1, %xmm0
; X86-SSE-NEXT: retl
%mod = and <16 x i8> %b, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7>
%splat = shufflevector <16 x i8> %mod, <16 x i8> undef, <16 x i32> zeroinitializer
%shift = shl <16 x i8> %a, %splat
ret <16 x i8> %shift
}
;
; Constant Shifts
;
define <2 x i64> @constant_shift_v2i64(<2 x i64> %a) nounwind {
; SSE2-LABEL: constant_shift_v2i64:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: psllq $1, %xmm1
; SSE2-NEXT: psllq $7, %xmm0
; SSE2-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE2-NEXT: retq
;
; SSE41-LABEL: constant_shift_v2i64:
; SSE41: # %bb.0:
; SSE41-NEXT: movdqa %xmm0, %xmm1
; SSE41-NEXT: psllq $7, %xmm1
; SSE41-NEXT: psllq $1, %xmm0
; SSE41-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT: retq
;
; AVX1-LABEL: constant_shift_v2i64:
; AVX1: # %bb.0:
; AVX1-NEXT: vpsllq $7, %xmm0, %xmm1
; AVX1-NEXT: vpsllq $1, %xmm0, %xmm0
; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
; AVX1-NEXT: retq
;
; AVX2-LABEL: constant_shift_v2i64:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: constant_shift_v2i64:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshlq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: constant_shift_v2i64:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: constant_shift_v2i64:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: constant_shift_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllvq {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: constant_shift_v2i64:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: movdqa %xmm0, %xmm1
; X86-SSE-NEXT: psllq $1, %xmm1
; X86-SSE-NEXT: psllq $7, %xmm0
; X86-SSE-NEXT: movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X86-SSE-NEXT: retl
%shift = shl <2 x i64> %a, <i64 1, i64 7>
ret <2 x i64> %shift
}
define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) nounwind {
; SSE2-LABEL: constant_shift_v4i32:
; SSE2: # %bb.0:
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = 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), %xmm1
; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE2-NEXT: retq
;
; SSE41-LABEL: constant_shift_v4i32:
; SSE41: # %bb.0:
; SSE41-NEXT: pmulld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: constant_shift_v4i32:
; AVX1: # %bb.0:
; AVX1-NEXT: vpmulld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: constant_shift_v4i32:
; AVX2: # %bb.0:
; AVX2-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX2-NEXT: retq
;
; XOPAVX1-LABEL: constant_shift_v4i32:
; XOPAVX1: # %bb.0:
; XOPAVX1-NEXT: vpshld {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX1-NEXT: retq
;
; XOPAVX2-LABEL: constant_shift_v4i32:
; XOPAVX2: # %bb.0:
; XOPAVX2-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: constant_shift_v4i32:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: constant_shift_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: constant_shift_v4i32:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; X86-SSE-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X86-SSE-NEXT: pmuludq {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X86-SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE-NEXT: retl
%shift = shl <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
ret <4 x i32> %shift
}
define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) nounwind {
; SSE-LABEL: constant_shift_v8i16:
; SSE: # %bb.0:
; SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: constant_shift_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: constant_shift_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpshlw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512DQ-LABEL: constant_shift_v8i16:
; AVX512DQ: # %bb.0:
; AVX512DQ-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v8i16:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512BW-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: constant_shift_v8i16:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: constant_shift_v8i16:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-NEXT: vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: constant_shift_v8i16:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
ret <8 x i16> %shift
}
define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) nounwind {
; SSE2-LABEL: constant_shift_v16i8:
; SSE2: # %bb.0:
; SSE2-NEXT: movdqa %xmm0, %xmm1
; SSE2-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
; SSE2-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE2-NEXT: movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; SSE2-NEXT: pand %xmm2, %xmm1
; SSE2-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE2-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE2-NEXT: pand %xmm2, %xmm0
; SSE2-NEXT: packuswb %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSE41-LABEL: constant_shift_v16i8:
; SSE41: # %bb.0:
; SSE41-NEXT: pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; SSE41-NEXT: punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
; SSE41-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE41-NEXT: movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; SSE41-NEXT: pand %xmm2, %xmm0
; SSE41-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
; SSE41-NEXT: pand %xmm2, %xmm1
; SSE41-NEXT: packuswb %xmm0, %xmm1
; SSE41-NEXT: movdqa %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: constant_shift_v16i8:
; AVX1: # %bb.0:
; AVX1-NEXT: vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
; AVX1-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; AVX1-NEXT: vpand %xmm2, %xmm1, %xmm1
; AVX1-NEXT: vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; AVX1-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: vpand %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vpackuswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: constant_shift_v16i8:
; AVX2: # %bb.0:
; AVX2-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
; AVX2-NEXT: vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT: vpackuswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; XOP-LABEL: constant_shift_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpshlb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512DQ-LABEL: constant_shift_v16i8:
; AVX512DQ: # %bb.0:
; AVX512DQ-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
; AVX512DQ-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQ-NEXT: vzeroupper
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v16i8:
; AVX512BW: # %bb.0:
; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,7,6,5,4,3,2,1,0]
; 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: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: # kill: def $xmm0 killed $xmm0 killed $ymm0
; AVX512BW-NEXT: vzeroupper
; AVX512BW-NEXT: retq
;
; AVX512DQVL-LABEL: constant_shift_v16i8:
; AVX512DQVL: # %bb.0:
; AVX512DQVL-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
; AVX512DQVL-NEXT: vpsllvd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0
; AVX512DQVL-NEXT: vzeroupper
; AVX512DQVL-NEXT: retq
;
; AVX512BWVL-LABEL: constant_shift_v16i8:
; AVX512BWVL: # %bb.0:
; AVX512BWVL-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
; AVX512BWVL-NEXT: vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0
; AVX512BWVL-NEXT: vzeroupper
; AVX512BWVL-NEXT: retq
;
; X86-SSE-LABEL: constant_shift_v16i8:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: movdqa %xmm0, %xmm1
; X86-SSE-NEXT: punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
; X86-SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
; X86-SSE-NEXT: movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
; X86-SSE-NEXT: pand %xmm2, %xmm1
; X86-SSE-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X86-SSE-NEXT: pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: pand %xmm2, %xmm0
; X86-SSE-NEXT: packuswb %xmm1, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>
ret <16 x i8> %shift
}
;
; Uniform Constant Shifts
;
define <2 x i64> @splatconstant_shift_v2i64(<2 x i64> %a) nounwind {
; SSE-LABEL: splatconstant_shift_v2i64:
; SSE: # %bb.0:
; SSE-NEXT: psllq $7, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_shift_v2i64:
; AVX: # %bb.0:
; AVX-NEXT: vpsllq $7, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatconstant_shift_v2i64:
; XOP: # %bb.0:
; XOP-NEXT: vpsllq $7, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatconstant_shift_v2i64:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllq $7, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_shift_v2i64:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllq $7, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatconstant_shift_v2i64:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: psllq $7, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <2 x i64> %a, <i64 7, i64 7>
ret <2 x i64> %shift
}
define <4 x i32> @splatconstant_shift_v4i32(<4 x i32> %a) nounwind {
; SSE-LABEL: splatconstant_shift_v4i32:
; SSE: # %bb.0:
; SSE-NEXT: pslld $5, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_shift_v4i32:
; AVX: # %bb.0:
; AVX-NEXT: vpslld $5, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatconstant_shift_v4i32:
; XOP: # %bb.0:
; XOP-NEXT: vpslld $5, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatconstant_shift_v4i32:
; AVX512: # %bb.0:
; AVX512-NEXT: vpslld $5, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_shift_v4i32:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpslld $5, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatconstant_shift_v4i32:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: pslld $5, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <4 x i32> %a, <i32 5, i32 5, i32 5, i32 5>
ret <4 x i32> %shift
}
define <8 x i16> @splatconstant_shift_v8i16(<8 x i16> %a) nounwind {
; SSE-LABEL: splatconstant_shift_v8i16:
; SSE: # %bb.0:
; SSE-NEXT: psllw $3, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_shift_v8i16:
; AVX: # %bb.0:
; AVX-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatconstant_shift_v8i16:
; XOP: # %bb.0:
; XOP-NEXT: vpsllw $3, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatconstant_shift_v8i16:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_shift_v8i16:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatconstant_shift_v8i16:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: psllw $3, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <8 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>
ret <8 x i16> %shift
}
define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) nounwind {
; SSE-LABEL: splatconstant_shift_v16i8:
; SSE: # %bb.0:
; SSE-NEXT: psllw $3, %xmm0
; SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: splatconstant_shift_v16i8:
; AVX: # %bb.0:
; AVX-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
;
; XOP-LABEL: splatconstant_shift_v16i8:
; XOP: # %bb.0:
; XOP-NEXT: vpshlb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: splatconstant_shift_v16i8:
; AVX512: # %bb.0:
; AVX512-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX512-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
;
; AVX512VL-LABEL: splatconstant_shift_v16i8:
; AVX512VL: # %bb.0:
; AVX512VL-NEXT: vpsllw $3, %xmm0, %xmm0
; AVX512VL-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT: retq
;
; X86-SSE-LABEL: splatconstant_shift_v16i8:
; X86-SSE: # %bb.0:
; X86-SSE-NEXT: psllw $3, %xmm0
; X86-SSE-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
; X86-SSE-NEXT: retl
%shift = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>
ret <16 x i8> %shift
}