; 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=AVX512,AVX512F
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vl | FileCheck %s --check-prefixes=AVX512,AVX512VL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=AVX512,AVX512BW
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512,AVX512VLBW
; 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 rotates.
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=X32-SSE,X32-SSE2

;
; Variable Rotates
;

define <2 x i64> @var_rotate_v2i64(<2 x i64> %a, <2 x i64> %b) nounwind {
; SSE2-LABEL: var_rotate_v2i64:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [64,64]
; SSE2-NEXT:    psubq %xmm1, %xmm2
; SSE2-NEXT:    movdqa %xmm0, %xmm3
; SSE2-NEXT:    psllq %xmm1, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
; SSE2-NEXT:    movdqa %xmm0, %xmm4
; SSE2-NEXT:    psllq %xmm1, %xmm4
; SSE2-NEXT:    movsd {{.*#+}} xmm4 = xmm3[0],xmm4[1]
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    psrlq %xmm2, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
; SSE2-NEXT:    psrlq %xmm2, %xmm0
; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE2-NEXT:    orpd %xmm4, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: var_rotate_v2i64:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [64,64]
; SSE41-NEXT:    psubq %xmm1, %xmm2
; SSE41-NEXT:    movdqa %xmm0, %xmm3
; SSE41-NEXT:    psllq %xmm1, %xmm3
; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
; SSE41-NEXT:    movdqa %xmm0, %xmm4
; SSE41-NEXT:    psllq %xmm1, %xmm4
; SSE41-NEXT:    pblendw {{.*#+}} xmm4 = xmm3[0,1,2,3],xmm4[4,5,6,7]
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    psrlq %xmm2, %xmm1
; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
; SSE41-NEXT:    psrlq %xmm2, %xmm0
; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
; SSE41-NEXT:    por %xmm4, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: var_rotate_v2i64:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [64,64]
; AVX1-NEXT:    vpsubq %xmm1, %xmm2, %xmm2
; AVX1-NEXT:    vpsllq %xmm1, %xmm0, %xmm3
; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
; AVX1-NEXT:    vpsllq %xmm1, %xmm0, %xmm1
; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm3[0,1,2,3],xmm1[4,5,6,7]
; AVX1-NEXT:    vpsrlq %xmm2, %xmm0, %xmm3
; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
; AVX1-NEXT:    vpsrlq %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm3[0,1,2,3],xmm0[4,5,6,7]
; AVX1-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: var_rotate_v2i64:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [64,64]
; AVX2-NEXT:    vpsubq %xmm1, %xmm2, %xmm2
; AVX2-NEXT:    vpsllvq %xmm1, %xmm0, %xmm1
; AVX2-NEXT:    vpsrlvq %xmm2, %xmm0, %xmm0
; AVX2-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: var_rotate_v2i64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: var_rotate_v2i64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolvq %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: var_rotate_v2i64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: var_rotate_v2i64:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolvq %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: var_rotate_v2i64:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotq %xmm1, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: var_rotate_v2i64:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [64,0,64,0]
; X32-SSE-NEXT:    psubq %xmm1, %xmm2
; X32-SSE-NEXT:    movdqa %xmm0, %xmm3
; X32-SSE-NEXT:    psllq %xmm1, %xmm3
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm4
; X32-SSE-NEXT:    psllq %xmm1, %xmm4
; X32-SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm3[0],xmm4[1]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlq %xmm2, %xmm1
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
; X32-SSE-NEXT:    psrlq %xmm2, %xmm0
; X32-SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X32-SSE-NEXT:    orpd %xmm4, %xmm0
; X32-SSE-NEXT:    retl
  %b64 = sub <2 x i64> <i64 64, i64 64>, %b
  %shl = shl <2 x i64> %a, %b
  %lshr = lshr <2 x i64> %a, %b64
  %or = or <2 x i64> %shl, %lshr
  ret <2 x i64> %or
}

define <4 x i32> @var_rotate_v4i32(<4 x i32> %a, <4 x i32> %b) nounwind {
; SSE2-LABEL: var_rotate_v4i32:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $23, %xmm1
; SSE2-NEXT:    paddd {{.*}}(%rip), %xmm1
; SSE2-NEXT:    cvttps2dq %xmm1, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; SSE2-NEXT:    pmuludq %xmm1, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,3,2,3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    pmuludq %xmm2, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,3,2,3]
; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; 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_rotate_v4i32:
; SSE41:       # %bb.0:
; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; SSE41-NEXT:    pslld $23, %xmm1
; SSE41-NEXT:    paddd {{.*}}(%rip), %xmm1
; SSE41-NEXT:    cvttps2dq %xmm1, %xmm1
; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,3,3]
; SSE41-NEXT:    pmuludq %xmm2, %xmm3
; SSE41-NEXT:    pmuludq %xmm1, %xmm0
; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,0,2,2]
; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; SSE41-NEXT:    por %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: var_rotate_v4i32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; AVX1-NEXT:    vpslld $23, %xmm1, %xmm1
; AVX1-NEXT:    vpaddd {{.*}}(%rip), %xmm1, %xmm1
; AVX1-NEXT:    vcvttps2dq %xmm1, %xmm1
; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm1[1,1,3,3]
; AVX1-NEXT:    vpmuludq %xmm3, %xmm2, %xmm2
; AVX1-NEXT:    vpmuludq %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,0,2,2]
; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: var_rotate_v4i32:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllvd %xmm1, %xmm0, %xmm2
; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm3 = [32,32,32,32]
; AVX2-NEXT:    vpsubd %xmm1, %xmm3, %xmm1
; AVX2-NEXT:    vpsrlvd %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: var_rotate_v4i32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: var_rotate_v4i32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolvd %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: var_rotate_v4i32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: var_rotate_v4i32:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolvd %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: var_rotate_v4i32:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotd %xmm1, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: var_rotate_v4i32:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    pslld $23, %xmm1
; X32-SSE-NEXT:    paddd {{\.LCPI.*}}, %xmm1
; X32-SSE-NEXT:    cvttps2dq %xmm1, %xmm1
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; X32-SSE-NEXT:    pmuludq %xmm1, %xmm0
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,3,2,3]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; X32-SSE-NEXT:    pmuludq %xmm2, %xmm1
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,3,2,3]
; X32-SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X32-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X32-SSE-NEXT:    por %xmm3, %xmm0
; X32-SSE-NEXT:    retl
  %b32 = sub <4 x i32> <i32 32, i32 32, i32 32, i32 32>, %b
  %shl = shl <4 x i32> %a, %b
  %lshr = lshr <4 x i32> %a, %b32
  %or = or <4 x i32> %shl, %lshr
  ret <4 x i32> %or
}

define <8 x i16> @var_rotate_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; SSE2-LABEL: var_rotate_v8i16:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pxor %xmm2, %xmm2
; SSE2-NEXT:    movdqa %xmm1, %xmm3
; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
; SSE2-NEXT:    pslld $23, %xmm3
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [1065353216,1065353216,1065353216,1065353216]
; SSE2-NEXT:    paddd %xmm4, %xmm3
; SSE2-NEXT:    cvttps2dq %xmm3, %xmm3
; SSE2-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,4,6,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
; SSE2-NEXT:    pslld $23, %xmm1
; SSE2-NEXT:    paddd %xmm4, %xmm1
; SSE2-NEXT:    cvttps2dq %xmm1, %xmm1
; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
; SSE2-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    pmulhuw %xmm1, %xmm2
; SSE2-NEXT:    pmullw %xmm1, %xmm0
; SSE2-NEXT:    por %xmm2, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: var_rotate_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],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[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:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    pmulhuw %xmm2, %xmm1
; SSE41-NEXT:    pmullw %xmm2, %xmm0
; SSE41-NEXT:    por %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: var_rotate_v8i16:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[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:    vpmulhuw %xmm1, %xmm0, %xmm2
; AVX1-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: var_rotate_v8i16:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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 %ymm2, %ymm0, %ymm2
; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [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 %ymm3, %ymm2, %ymm2
; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = [16,16,16,16,16,16,16,16]
; AVX2-NEXT:    vpsubw %xmm1, %xmm4, %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:    vpsrlvd %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
; AVX2-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: var_rotate_v8i16:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpsllvd %ymm2, %ymm0, %ymm2
; AVX512F-NEXT:    vpmovdw %zmm2, %ymm2
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX512F-NEXT:    vpsubw %xmm1, %xmm3, %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:    vpsrlvd %ymm1, %ymm0, %ymm0
; AVX512F-NEXT:    vpmovdw %zmm0, %ymm0
; AVX512F-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: var_rotate_v8i16:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpsllvd %ymm2, %ymm0, %ymm2
; AVX512VL-NEXT:    vpmovdw %ymm2, %xmm2
; AVX512VL-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX512VL-NEXT:    vpsubw %xmm1, %xmm3, %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:    vpsrlvd %ymm1, %ymm0, %ymm0
; AVX512VL-NEXT:    vpmovdw %ymm0, %xmm0
; AVX512VL-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512VL-NEXT:    vzeroupper
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: var_rotate_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, %zmm2
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX512BW-NEXT:    vpsubw %xmm1, %xmm3, %xmm1
; AVX512BW-NEXT:    vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: var_rotate_v8i16:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpsllvw %xmm1, %xmm0, %xmm2
; AVX512VLBW-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX512VLBW-NEXT:    vpsubw %xmm1, %xmm3, %xmm1
; AVX512VLBW-NEXT:    vpsrlvw %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: var_rotate_v8i16:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotw %xmm1, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: var_rotate_v8i16:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    pxor %xmm2, %xmm2
; X32-SSE-NEXT:    movdqa %xmm1, %xmm3
; X32-SSE-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm2[4],xmm3[5],xmm2[5],xmm3[6],xmm2[6],xmm3[7],xmm2[7]
; X32-SSE-NEXT:    pslld $23, %xmm3
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm4 = [1065353216,1065353216,1065353216,1065353216]
; X32-SSE-NEXT:    paddd %xmm4, %xmm3
; X32-SSE-NEXT:    cvttps2dq %xmm3, %xmm3
; X32-SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
; X32-SSE-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,4,6,6,7]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
; X32-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
; X32-SSE-NEXT:    pslld $23, %xmm1
; X32-SSE-NEXT:    paddd %xmm4, %xmm1
; X32-SSE-NEXT:    cvttps2dq %xmm1, %xmm1
; X32-SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
; X32-SSE-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X32-SSE-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    pmulhuw %xmm1, %xmm2
; X32-SSE-NEXT:    pmullw %xmm1, %xmm0
; X32-SSE-NEXT:    por %xmm2, %xmm0
; X32-SSE-NEXT:    retl
  %b16 = sub <8 x i16> <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16>, %b
  %shl = shl <8 x i16> %a, %b
  %lshr = lshr <8 x i16> %a, %b16
  %or = or <8 x i16> %shl, %lshr
  ret <8 x i16> %or
}

define <16 x i8> @var_rotate_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind {
; SSE2-LABEL: var_rotate_v16i8:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    pxor %xmm3, %xmm3
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm3
; SSE2-NEXT:    movdqa %xmm2, %xmm4
; SSE2-NEXT:    psrlw $4, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    movdqa %xmm2, %xmm5
; SSE2-NEXT:    psllw $4, %xmm5
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm5
; SSE2-NEXT:    por %xmm4, %xmm5
; SSE2-NEXT:    pand %xmm3, %xmm5
; SSE2-NEXT:    pandn %xmm2, %xmm3
; SSE2-NEXT:    por %xmm5, %xmm3
; SSE2-NEXT:    movdqa %xmm3, %xmm2
; SSE2-NEXT:    psrlw $6, %xmm2
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    psllw $2, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    por %xmm2, %xmm4
; SSE2-NEXT:    paddb %xmm1, %xmm1
; SSE2-NEXT:    pxor %xmm2, %xmm2
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm2
; SSE2-NEXT:    pand %xmm2, %xmm4
; SSE2-NEXT:    pandn %xmm3, %xmm2
; SSE2-NEXT:    por %xmm4, %xmm2
; SSE2-NEXT:    movdqa %xmm2, %xmm3
; SSE2-NEXT:    paddb %xmm2, %xmm3
; SSE2-NEXT:    movdqa %xmm2, %xmm4
; SSE2-NEXT:    psrlw $7, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    por %xmm3, %xmm4
; SSE2-NEXT:    paddb %xmm1, %xmm1
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm0
; SSE2-NEXT:    pand %xmm0, %xmm4
; SSE2-NEXT:    pandn %xmm2, %xmm0
; SSE2-NEXT:    por %xmm4, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: var_rotate_v16i8:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa %xmm1, %xmm2
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    psrlw $4, %xmm0
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psllw $4, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    psllw $5, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    psrlw $6, %xmm0
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psllw $2, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    paddb %xmm2, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    paddb %xmm1, %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psrlw $7, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    paddb %xmm2, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX-LABEL: var_rotate_v16i8:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: var_rotate_v16i8:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512F-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512F-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX512F-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX512F-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: var_rotate_v16i8:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512VL-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512VL-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX512VL-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: var_rotate_v16i8:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} ymm2 = 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 %zmm2, %zmm0, %zmm2
; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
; AVX512BW-NEXT:    vpsubb %xmm1, %xmm3, %xmm1
; 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:    vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: var_rotate_v16i8:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpmovzxbw {{.*#+}} ymm2 = 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:    vpsllvw %ymm2, %ymm0, %ymm2
; AVX512VLBW-NEXT:    vpmovwb %ymm2, %xmm2
; AVX512VLBW-NEXT:    vmovdqa {{.*#+}} xmm3 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
; AVX512VLBW-NEXT:    vpsubb %xmm1, %xmm3, %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:    vpsrlvw %ymm1, %ymm0, %ymm0
; AVX512VLBW-NEXT:    vpmovwb %ymm0, %xmm0
; AVX512VLBW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512VLBW-NEXT:    vzeroupper
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: var_rotate_v16i8:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotb %xmm1, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: var_rotate_v16i8:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    psllw $5, %xmm1
; X32-SSE-NEXT:    pxor %xmm0, %xmm0
; X32-SSE-NEXT:    pxor %xmm3, %xmm3
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm3
; X32-SSE-NEXT:    movdqa %xmm2, %xmm4
; X32-SSE-NEXT:    psrlw $4, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    movdqa %xmm2, %xmm5
; X32-SSE-NEXT:    psllw $4, %xmm5
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm5
; X32-SSE-NEXT:    por %xmm4, %xmm5
; X32-SSE-NEXT:    pand %xmm3, %xmm5
; X32-SSE-NEXT:    pandn %xmm2, %xmm3
; X32-SSE-NEXT:    por %xmm5, %xmm3
; X32-SSE-NEXT:    movdqa %xmm3, %xmm2
; X32-SSE-NEXT:    psrlw $6, %xmm2
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm2
; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
; X32-SSE-NEXT:    psllw $2, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    por %xmm2, %xmm4
; X32-SSE-NEXT:    paddb %xmm1, %xmm1
; X32-SSE-NEXT:    pxor %xmm2, %xmm2
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm2
; X32-SSE-NEXT:    pand %xmm2, %xmm4
; X32-SSE-NEXT:    pandn %xmm3, %xmm2
; X32-SSE-NEXT:    por %xmm4, %xmm2
; X32-SSE-NEXT:    movdqa %xmm2, %xmm3
; X32-SSE-NEXT:    paddb %xmm2, %xmm3
; X32-SSE-NEXT:    movdqa %xmm2, %xmm4
; X32-SSE-NEXT:    psrlw $7, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    por %xmm3, %xmm4
; X32-SSE-NEXT:    paddb %xmm1, %xmm1
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm0
; X32-SSE-NEXT:    pand %xmm0, %xmm4
; X32-SSE-NEXT:    pandn %xmm2, %xmm0
; X32-SSE-NEXT:    por %xmm4, %xmm0
; X32-SSE-NEXT:    retl
  %b8 = sub <16 x i8> <i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8>, %b
  %shl = shl <16 x i8> %a, %b
  %lshr = lshr <16 x i8> %a, %b8
  %or = or <16 x i8> %shl, %lshr
  ret <16 x i8> %or
}

;
; Uniform Variable Rotates
;

define <2 x i64> @splatvar_rotate_v2i64(<2 x i64> %a, <2 x i64> %b) nounwind {
; SSE-LABEL: splatvar_rotate_v2i64:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [64,64]
; SSE-NEXT:    psubq %xmm1, %xmm2
; SSE-NEXT:    movdqa %xmm0, %xmm3
; SSE-NEXT:    psllq %xmm1, %xmm3
; SSE-NEXT:    psrlq %xmm2, %xmm0
; SSE-NEXT:    por %xmm3, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatvar_rotate_v2i64:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovdqa {{.*#+}} xmm2 = [64,64]
; AVX-NEXT:    vpsubq %xmm1, %xmm2, %xmm2
; AVX-NEXT:    vpsllq %xmm1, %xmm0, %xmm1
; AVX-NEXT:    vpsrlq %xmm2, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: splatvar_rotate_v2i64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vpbroadcastq %xmm1, %xmm1
; AVX512F-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatvar_rotate_v2i64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpbroadcastq %xmm1, %xmm1
; AVX512VL-NEXT:    vprolvq %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatvar_rotate_v2i64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vpbroadcastq %xmm1, %xmm1
; AVX512BW-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatvar_rotate_v2i64:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpbroadcastq %xmm1, %xmm1
; AVX512VLBW-NEXT:    vprolvq %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOPAVX1-LABEL: splatvar_rotate_v2i64:
; XOPAVX1:       # %bb.0:
; XOPAVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
; XOPAVX1-NEXT:    vprotq %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT:    retq
;
; XOPAVX2-LABEL: splatvar_rotate_v2i64:
; XOPAVX2:       # %bb.0:
; XOPAVX2-NEXT:    vpbroadcastq %xmm1, %xmm1
; XOPAVX2-NEXT:    vprotq %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT:    retq
;
; X32-SSE-LABEL: splatvar_rotate_v2i64:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[0,1,0,1]
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [64,0,64,0]
; X32-SSE-NEXT:    psubq %xmm2, %xmm3
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    psllq %xmm1, %xmm2
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlq %xmm3, %xmm1
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[2,3,0,1]
; X32-SSE-NEXT:    psrlq %xmm3, %xmm0
; X32-SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X32-SSE-NEXT:    orpd %xmm2, %xmm0
; X32-SSE-NEXT:    retl
  %splat = shufflevector <2 x i64> %b, <2 x i64> undef, <2 x i32> zeroinitializer
  %splat64 = sub <2 x i64> <i64 64, i64 64>, %splat
  %shl = shl <2 x i64> %a, %splat
  %lshr = lshr <2 x i64> %a, %splat64
  %or = or <2 x i64> %shl, %lshr
  ret <2 x i64> %or
}

define <4 x i32> @splatvar_rotate_v4i32(<4 x i32> %a, <4 x i32> %b) nounwind {
; SSE2-LABEL: splatvar_rotate_v4i32:
; SSE2:       # %bb.0:
; SSE2-NEXT:    xorps %xmm2, %xmm2
; SSE2-NEXT:    xorps %xmm3, %xmm3
; SSE2-NEXT:    movss {{.*#+}} xmm3 = xmm1[0],xmm3[1,2,3]
; SSE2-NEXT:    movdqa %xmm0, %xmm4
; SSE2-NEXT:    pslld %xmm3, %xmm4
; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [32,32,32,32]
; SSE2-NEXT:    psubd %xmm1, %xmm3
; SSE2-NEXT:    movss {{.*#+}} xmm2 = xmm3[0],xmm2[1,2,3]
; SSE2-NEXT:    psrld %xmm2, %xmm0
; SSE2-NEXT:    por %xmm4, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: splatvar_rotate_v4i32:
; SSE41:       # %bb.0:
; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero
; SSE41-NEXT:    movdqa %xmm0, %xmm3
; SSE41-NEXT:    pslld %xmm2, %xmm3
; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [32,32,32,32]
; SSE41-NEXT:    psubd %xmm1, %xmm2
; SSE41-NEXT:    pmovzxdq {{.*#+}} xmm1 = xmm2[0],zero,xmm2[1],zero
; SSE41-NEXT:    psrld %xmm1, %xmm0
; SSE41-NEXT:    por %xmm3, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: splatvar_rotate_v4i32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero
; AVX1-NEXT:    vpslld %xmm2, %xmm0, %xmm2
; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [32,32,32,32]
; AVX1-NEXT:    vpsubd %xmm1, %xmm3, %xmm1
; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; AVX1-NEXT:    vpsrld %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: splatvar_rotate_v4i32:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovzxdq {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero
; AVX2-NEXT:    vpslld %xmm2, %xmm0, %xmm2
; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm3 = [32,32,32,32]
; AVX2-NEXT:    vpsubd %xmm1, %xmm3, %xmm1
; AVX2-NEXT:    vpmovzxdq {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero
; AVX2-NEXT:    vpsrld %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: splatvar_rotate_v4i32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vpbroadcastd %xmm1, %xmm1
; AVX512F-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatvar_rotate_v4i32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpbroadcastd %xmm1, %xmm1
; AVX512VL-NEXT:    vprolvd %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatvar_rotate_v4i32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vpbroadcastd %xmm1, %xmm1
; AVX512BW-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatvar_rotate_v4i32:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpbroadcastd %xmm1, %xmm1
; AVX512VLBW-NEXT:    vprolvd %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOPAVX1-LABEL: splatvar_rotate_v4i32:
; XOPAVX1:       # %bb.0:
; XOPAVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; XOPAVX1-NEXT:    vprotd %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT:    retq
;
; XOPAVX2-LABEL: splatvar_rotate_v4i32:
; XOPAVX2:       # %bb.0:
; XOPAVX2-NEXT:    vpbroadcastd %xmm1, %xmm1
; XOPAVX2-NEXT:    vprotd %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT:    retq
;
; X32-SSE-LABEL: splatvar_rotate_v4i32:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    xorps %xmm2, %xmm2
; X32-SSE-NEXT:    xorps %xmm3, %xmm3
; X32-SSE-NEXT:    movss {{.*#+}} xmm3 = xmm1[0],xmm3[1,2,3]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm4
; X32-SSE-NEXT:    pslld %xmm3, %xmm4
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm3 = [32,32,32,32]
; X32-SSE-NEXT:    psubd %xmm1, %xmm3
; X32-SSE-NEXT:    movss {{.*#+}} xmm2 = xmm3[0],xmm2[1,2,3]
; X32-SSE-NEXT:    psrld %xmm2, %xmm0
; X32-SSE-NEXT:    por %xmm4, %xmm0
; X32-SSE-NEXT:    retl
  %splat = shufflevector <4 x i32> %b, <4 x i32> undef, <4 x i32> zeroinitializer
  %splat32 = sub <4 x i32> <i32 32, i32 32, i32 32, i32 32>, %splat
  %shl = shl <4 x i32> %a, %splat
  %lshr = lshr <4 x i32> %a, %splat32
  %or = or <4 x i32> %shl, %lshr
  ret <4 x i32> %or
}

define <8 x i16> @splatvar_rotate_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; SSE2-LABEL: splatvar_rotate_v8i16:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [16,16,16,16,16,16,16,16]
; SSE2-NEXT:    psubw %xmm1, %xmm2
; 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:    movdqa %xmm0, %xmm3
; SSE2-NEXT:    psllw %xmm1, %xmm3
; SSE2-NEXT:    pslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm2[0,1]
; SSE2-NEXT:    psrldq {{.*#+}} xmm2 = xmm2[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSE2-NEXT:    psrlw %xmm2, %xmm0
; SSE2-NEXT:    por %xmm3, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: splatvar_rotate_v8i16:
; SSE41:       # %bb.0:
; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; SSE41-NEXT:    movdqa %xmm0, %xmm3
; SSE41-NEXT:    psllw %xmm2, %xmm3
; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [16,16,16,16,16,16,16,16]
; SSE41-NEXT:    psubw %xmm1, %xmm2
; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm1 = xmm2[0],zero,zero,zero,xmm2[1],zero,zero,zero
; SSE41-NEXT:    psrlw %xmm1, %xmm0
; SSE41-NEXT:    por %xmm3, %xmm0
; SSE41-NEXT:    retq
;
; AVX-LABEL: splatvar_rotate_v8i16:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovzxwq {{.*#+}} xmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX-NEXT:    vpsllw %xmm2, %xmm0, %xmm2
; AVX-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX-NEXT:    vpsubw %xmm1, %xmm3, %xmm1
; AVX-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX-NEXT:    vpsrlw %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX-NEXT:    retq
;
; AVX512-LABEL: splatvar_rotate_v8i16:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm2 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX512-NEXT:    vpsllw %xmm2, %xmm0, %xmm2
; AVX512-NEXT:    vmovdqa {{.*#+}} xmm3 = [16,16,16,16,16,16,16,16]
; AVX512-NEXT:    vpsubw %xmm1, %xmm3, %xmm1
; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
; AVX512-NEXT:    vpsrlw %xmm1, %xmm0, %xmm0
; AVX512-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512-NEXT:    retq
;
; XOPAVX1-LABEL: splatvar_rotate_v8i16:
; XOPAVX1:       # %bb.0:
; XOPAVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,2,3,4,5,6,7]
; XOPAVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
; XOPAVX1-NEXT:    vprotw %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT:    retq
;
; XOPAVX2-LABEL: splatvar_rotate_v8i16:
; XOPAVX2:       # %bb.0:
; XOPAVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
; XOPAVX2-NEXT:    vprotw %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT:    retq
;
; X32-SSE-LABEL: splatvar_rotate_v8i16:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [16,16,16,16,16,16,16,16]
; X32-SSE-NEXT:    psubw %xmm1, %xmm2
; X32-SSE-NEXT:    pslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
; X32-SSE-NEXT:    psrldq {{.*#+}} xmm1 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; X32-SSE-NEXT:    movdqa %xmm0, %xmm3
; X32-SSE-NEXT:    psllw %xmm1, %xmm3
; X32-SSE-NEXT:    pslldq {{.*#+}} xmm2 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm2[0,1]
; X32-SSE-NEXT:    psrldq {{.*#+}} xmm2 = xmm2[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; X32-SSE-NEXT:    psrlw %xmm2, %xmm0
; X32-SSE-NEXT:    por %xmm3, %xmm0
; X32-SSE-NEXT:    retl
  %splat = shufflevector <8 x i16> %b, <8 x i16> undef, <8 x i32> zeroinitializer
  %splat16 = sub <8 x i16> <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16>, %splat
  %shl = shl <8 x i16> %a, %splat
  %lshr = lshr <8 x i16> %a, %splat16
  %or = or <8 x i16> %shl, %lshr
  ret <8 x i16> %or
}

define <16 x i8> @splatvar_rotate_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind {
; SSE2-LABEL: splatvar_rotate_v16i8:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm1[0,0,2,3,4,5,6,7]
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,0,0,0]
; SSE2-NEXT:    movdqa %xmm2, %xmm0
; SSE2-NEXT:    psrlw $4, %xmm0
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE2-NEXT:    movdqa %xmm2, %xmm3
; SSE2-NEXT:    psllw $4, %xmm3
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE2-NEXT:    por %xmm0, %xmm3
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    pxor %xmm4, %xmm4
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm4
; SSE2-NEXT:    pand %xmm4, %xmm3
; SSE2-NEXT:    pandn %xmm2, %xmm4
; SSE2-NEXT:    por %xmm3, %xmm4
; SSE2-NEXT:    movdqa %xmm4, %xmm2
; SSE2-NEXT:    psrlw $6, %xmm2
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    psllw $2, %xmm3
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE2-NEXT:    por %xmm2, %xmm3
; SSE2-NEXT:    paddb %xmm1, %xmm1
; SSE2-NEXT:    pxor %xmm2, %xmm2
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm2
; SSE2-NEXT:    pand %xmm2, %xmm3
; SSE2-NEXT:    pandn %xmm4, %xmm2
; SSE2-NEXT:    por %xmm3, %xmm2
; SSE2-NEXT:    movdqa %xmm2, %xmm3
; SSE2-NEXT:    paddb %xmm2, %xmm3
; SSE2-NEXT:    movdqa %xmm2, %xmm4
; SSE2-NEXT:    psrlw $7, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    por %xmm3, %xmm4
; SSE2-NEXT:    paddb %xmm1, %xmm1
; SSE2-NEXT:    pcmpgtb %xmm1, %xmm0
; SSE2-NEXT:    pand %xmm0, %xmm4
; SSE2-NEXT:    pandn %xmm2, %xmm0
; SSE2-NEXT:    por %xmm4, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: splatvar_rotate_v16i8:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa %xmm1, %xmm2
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    pxor %xmm0, %xmm0
; SSE41-NEXT:    pshufb %xmm0, %xmm2
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    psrlw $4, %xmm0
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psllw $4, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    psllw $5, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    psrlw $6, %xmm0
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psllw $2, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    paddb %xmm2, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    paddb %xmm1, %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psrlw $7, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm0, %xmm3
; SSE41-NEXT:    paddb %xmm2, %xmm2
; SSE41-NEXT:    movdqa %xmm2, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: splatvar_rotate_v16i8:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX1-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX1-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX1-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX1-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX1-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX1-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX1-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX1-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: splatvar_rotate_v16i8:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpbroadcastb %xmm1, %xmm1
; AVX2-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX2-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX2-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX2-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX2-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX2-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX2-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX2-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX2-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX2-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX2-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX2-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX2-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX2-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX2-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: splatvar_rotate_v16i8:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpbroadcastb %xmm1, %xmm1
; AVX512F-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512F-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512F-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX512F-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX512F-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512F-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatvar_rotate_v16i8:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpbroadcastb %xmm1, %xmm1
; AVX512VL-NEXT:    vpsrlw $4, %xmm0, %xmm2
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT:    vpsllw $4, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512VL-NEXT:    vpsllw $5, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    vpsrlw $6, %xmm0, %xmm2
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm2, %xmm3, %xmm2
; AVX512VL-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    vpaddb %xmm0, %xmm0, %xmm2
; AVX512VL-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm3, %xmm2, %xmm2
; AVX512VL-NEXT:    vpaddb %xmm1, %xmm1, %xmm1
; AVX512VL-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatvar_rotate_v16i8:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vpbroadcastb %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:    vpmovzxbw {{.*#+}} ymm2 = 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:    vpsllvw %zmm2, %zmm0, %zmm2
; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
; AVX512BW-NEXT:    vpsubb %xmm1, %xmm3, %xmm1
; 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:    vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatvar_rotate_v16i8:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpbroadcastb %xmm1, %xmm1
; 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:    vpmovzxbw {{.*#+}} ymm2 = 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 %ymm2, %ymm0, %ymm2
; AVX512VLBW-NEXT:    vpmovwb %ymm2, %xmm2
; AVX512VLBW-NEXT:    vmovdqa {{.*#+}} xmm3 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
; AVX512VLBW-NEXT:    vpsubb %xmm1, %xmm3, %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:    vpsrlvw %ymm1, %ymm0, %ymm0
; AVX512VLBW-NEXT:    vpmovwb %ymm0, %xmm0
; AVX512VLBW-NEXT:    vpor %xmm0, %xmm2, %xmm0
; AVX512VLBW-NEXT:    vzeroupper
; AVX512VLBW-NEXT:    retq
;
; XOPAVX1-LABEL: splatvar_rotate_v16i8:
; XOPAVX1:       # %bb.0:
; XOPAVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; XOPAVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
; XOPAVX1-NEXT:    vprotb %xmm1, %xmm0, %xmm0
; XOPAVX1-NEXT:    retq
;
; XOPAVX2-LABEL: splatvar_rotate_v16i8:
; XOPAVX2:       # %bb.0:
; XOPAVX2-NEXT:    vpbroadcastb %xmm1, %xmm1
; XOPAVX2-NEXT:    vprotb %xmm1, %xmm0, %xmm0
; XOPAVX2-NEXT:    retq
;
; X32-SSE-LABEL: splatvar_rotate_v16i8:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; X32-SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm1[0,0,2,3,4,5,6,7]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,0,0,0]
; X32-SSE-NEXT:    movdqa %xmm2, %xmm0
; X32-SSE-NEXT:    psrlw $4, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    movdqa %xmm2, %xmm3
; X32-SSE-NEXT:    psllw $4, %xmm3
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm3
; X32-SSE-NEXT:    por %xmm0, %xmm3
; X32-SSE-NEXT:    psllw $5, %xmm1
; X32-SSE-NEXT:    pxor %xmm0, %xmm0
; X32-SSE-NEXT:    pxor %xmm4, %xmm4
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm4
; X32-SSE-NEXT:    pand %xmm4, %xmm3
; X32-SSE-NEXT:    pandn %xmm2, %xmm4
; X32-SSE-NEXT:    por %xmm3, %xmm4
; X32-SSE-NEXT:    movdqa %xmm4, %xmm2
; X32-SSE-NEXT:    psrlw $6, %xmm2
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm2
; X32-SSE-NEXT:    movdqa %xmm4, %xmm3
; X32-SSE-NEXT:    psllw $2, %xmm3
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm3
; X32-SSE-NEXT:    por %xmm2, %xmm3
; X32-SSE-NEXT:    paddb %xmm1, %xmm1
; X32-SSE-NEXT:    pxor %xmm2, %xmm2
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm2
; X32-SSE-NEXT:    pand %xmm2, %xmm3
; X32-SSE-NEXT:    pandn %xmm4, %xmm2
; X32-SSE-NEXT:    por %xmm3, %xmm2
; X32-SSE-NEXT:    movdqa %xmm2, %xmm3
; X32-SSE-NEXT:    paddb %xmm2, %xmm3
; X32-SSE-NEXT:    movdqa %xmm2, %xmm4
; X32-SSE-NEXT:    psrlw $7, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    por %xmm3, %xmm4
; X32-SSE-NEXT:    paddb %xmm1, %xmm1
; X32-SSE-NEXT:    pcmpgtb %xmm1, %xmm0
; X32-SSE-NEXT:    pand %xmm0, %xmm4
; X32-SSE-NEXT:    pandn %xmm2, %xmm0
; X32-SSE-NEXT:    por %xmm4, %xmm0
; X32-SSE-NEXT:    retl
  %splat = shufflevector <16 x i8> %b, <16 x i8> undef, <16 x i32> zeroinitializer
  %splat8 = sub <16 x i8> <i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8>, %splat
  %shl = shl <16 x i8> %a, %splat
  %lshr = lshr <16 x i8> %a, %splat8
  %or = or <16 x i8> %shl, %lshr
  ret <16 x i8> %or
}

;
; Constant Rotates
;

define <2 x i64> @constant_rotate_v2i64(<2 x i64> %a) nounwind {
; SSE2-LABEL: constant_rotate_v2i64:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    psllq $4, %xmm1
; SSE2-NEXT:    movdqa %xmm0, %xmm2
; SSE2-NEXT:    psllq $14, %xmm2
; SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm1[0],xmm2[1]
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    psrlq $60, %xmm1
; SSE2-NEXT:    psrlq $50, %xmm0
; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; SSE2-NEXT:    orpd %xmm2, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: constant_rotate_v2i64:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    psllq $14, %xmm1
; SSE41-NEXT:    movdqa %xmm0, %xmm2
; SSE41-NEXT:    psllq $4, %xmm2
; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm1[4,5,6,7]
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    psrlq $50, %xmm1
; SSE41-NEXT:    psrlq $60, %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_rotate_v2i64:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllq $14, %xmm0, %xmm1
; AVX1-NEXT:    vpsllq $4, %xmm0, %xmm2
; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
; AVX1-NEXT:    vpsrlq $50, %xmm0, %xmm2
; AVX1-NEXT:    vpsrlq $60, %xmm0, %xmm0
; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
; AVX1-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: constant_rotate_v2i64:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllvq {{.*}}(%rip), %xmm0, %xmm1
; AVX2-NEXT:    vpsrlvq {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: constant_rotate_v2i64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,14]
; AVX512F-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: constant_rotate_v2i64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolvq {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: constant_rotate_v2i64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,14]
; AVX512BW-NEXT:    vprolvq %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: constant_rotate_v2i64:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolvq {{.*}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: constant_rotate_v2i64:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotq {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: constant_rotate_v2i64:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psllq $4, %xmm1
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    psllq $14, %xmm2
; X32-SSE-NEXT:    movsd {{.*#+}} xmm2 = xmm1[0],xmm2[1]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlq $60, %xmm1
; X32-SSE-NEXT:    psrlq $50, %xmm0
; X32-SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
; X32-SSE-NEXT:    orpd %xmm2, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <2 x i64> %a, <i64 4, i64 14>
  %lshr = lshr <2 x i64> %a, <i64 60, i64 50>
  %or = or <2 x i64> %shl, %lshr
  ret <2 x i64> %or
}

define <4 x i32> @constant_rotate_v4i32(<4 x i32> %a) nounwind {
; SSE2-LABEL: constant_rotate_v4i32:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [16,32,64,128]
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; SSE2-NEXT:    pmuludq %xmm1, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,3,2,3]
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    pmuludq %xmm2, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,3,2,3]
; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; 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: constant_rotate_v4i32:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [16,32,64,128]
; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
; SSE41-NEXT:    pmuludq %xmm2, %xmm3
; SSE41-NEXT:    pmuludq %xmm1, %xmm0
; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,0,2,2]
; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; SSE41-NEXT:    por %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX1-LABEL: constant_rotate_v4i32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [16,32,64,128]
; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
; AVX1-NEXT:    vpmuludq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpmuludq %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm2[2,3],xmm1[4,5],xmm2[6,7]
; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,0,2,2]
; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    retq
;
; AVX2-LABEL: constant_rotate_v4i32:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsrlvd {{.*}}(%rip), %xmm0, %xmm1
; AVX2-NEXT:    vpsllvd {{.*}}(%rip), %xmm0, %xmm0
; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    retq
;
; AVX512F-LABEL: constant_rotate_v4i32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,5,6,7]
; AVX512F-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: constant_rotate_v4i32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolvd {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: constant_rotate_v4i32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,5,6,7]
; AVX512BW-NEXT:    vprolvd %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: constant_rotate_v4i32:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolvd {{.*}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: constant_rotate_v4i32:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotd {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: constant_rotate_v4i32:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm1 = [16,32,64,128]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
; X32-SSE-NEXT:    pmuludq %xmm1, %xmm0
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,3,2,3]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; X32-SSE-NEXT:    pmuludq %xmm2, %xmm1
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,3,2,3]
; X32-SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; X32-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
; X32-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X32-SSE-NEXT:    por %xmm3, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
  %lshr = lshr <4 x i32> %a, <i32 28, i32 27, i32 26, i32 25>
  %or = or <4 x i32> %shl, %lshr
  ret <4 x i32> %or
}

define <8 x i16> @constant_rotate_v8i16(<8 x i16> %a) nounwind {
; SSE-LABEL: constant_rotate_v8i16:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128]
; SSE-NEXT:    movdqa %xmm0, %xmm2
; SSE-NEXT:    pmulhuw %xmm1, %xmm2
; SSE-NEXT:    pmullw %xmm1, %xmm0
; SSE-NEXT:    por %xmm2, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: constant_rotate_v8i16:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128]
; AVX-NEXT:    vpmulhuw %xmm1, %xmm0, %xmm2
; AVX-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: constant_rotate_v8i16:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128]
; AVX512F-NEXT:    vpmulhuw %xmm1, %xmm0, %xmm2
; AVX512F-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: constant_rotate_v8i16:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128]
; AVX512VL-NEXT:    vpmulhuw %xmm1, %xmm0, %xmm2
; AVX512VL-NEXT:    vpmullw %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: constant_rotate_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:    vmovdqa {{.*#+}} xmm2 = [16,15,14,13,12,11,10,9]
; AVX512BW-NEXT:    vpsrlvw %zmm2, %zmm0, %zmm2
; AVX512BW-NEXT:    vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT:    vpor %xmm2, %xmm0, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: constant_rotate_v8i16:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vpsrlvw {{.*}}(%rip), %xmm0, %xmm1
; AVX512VLBW-NEXT:    vpsllvw {{.*}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: constant_rotate_v8i16:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotw {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: constant_rotate_v8i16:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm1 = [1,2,4,8,16,32,64,128]
; X32-SSE-NEXT:    movdqa %xmm0, %xmm2
; X32-SSE-NEXT:    pmulhuw %xmm1, %xmm2
; X32-SSE-NEXT:    pmullw %xmm1, %xmm0
; X32-SSE-NEXT:    por %xmm2, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <8 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>
  %lshr = lshr <8 x i16> %a, <i16 16, i16 15, i16 14, i16 13, i16 12, i16 11, i16 10, i16 9>
  %or = or <8 x i16> %shl, %lshr
  ret <8 x i16> %or
}

define <16 x i8> @constant_rotate_v16i8(<16 x i8> %a) nounwind {
; SSE2-LABEL: constant_rotate_v16i8:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa %xmm0, %xmm1
; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,57600,41152,24704,8256]
; SSE2-NEXT:    pxor %xmm0, %xmm0
; SSE2-NEXT:    pxor %xmm3, %xmm3
; SSE2-NEXT:    pcmpgtb %xmm2, %xmm3
; SSE2-NEXT:    movdqa %xmm1, %xmm4
; SSE2-NEXT:    psrlw $4, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    movdqa %xmm1, %xmm5
; SSE2-NEXT:    psllw $4, %xmm5
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm5
; SSE2-NEXT:    por %xmm4, %xmm5
; SSE2-NEXT:    pand %xmm3, %xmm5
; SSE2-NEXT:    pandn %xmm1, %xmm3
; SSE2-NEXT:    por %xmm5, %xmm3
; SSE2-NEXT:    movdqa %xmm3, %xmm1
; SSE2-NEXT:    psrlw $6, %xmm1
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
; SSE2-NEXT:    movdqa %xmm3, %xmm4
; SSE2-NEXT:    psllw $2, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    por %xmm1, %xmm4
; SSE2-NEXT:    paddb %xmm2, %xmm2
; SSE2-NEXT:    pxor %xmm1, %xmm1
; SSE2-NEXT:    pcmpgtb %xmm2, %xmm1
; SSE2-NEXT:    pand %xmm1, %xmm4
; SSE2-NEXT:    pandn %xmm3, %xmm1
; SSE2-NEXT:    por %xmm4, %xmm1
; SSE2-NEXT:    movdqa %xmm1, %xmm3
; SSE2-NEXT:    paddb %xmm1, %xmm3
; SSE2-NEXT:    movdqa %xmm1, %xmm4
; SSE2-NEXT:    psrlw $7, %xmm4
; SSE2-NEXT:    pand {{.*}}(%rip), %xmm4
; SSE2-NEXT:    por %xmm3, %xmm4
; SSE2-NEXT:    paddb %xmm2, %xmm2
; SSE2-NEXT:    pcmpgtb %xmm2, %xmm0
; SSE2-NEXT:    pand %xmm0, %xmm4
; SSE2-NEXT:    pandn %xmm1, %xmm0
; SSE2-NEXT:    por %xmm4, %xmm0
; SSE2-NEXT:    retq
;
; SSE41-LABEL: constant_rotate_v16i8:
; SSE41:       # %bb.0:
; SSE41-NEXT:    movdqa %xmm0, %xmm1
; SSE41-NEXT:    psrlw $4, %xmm0
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE41-NEXT:    movdqa %xmm1, %xmm2
; SSE41-NEXT:    psllw $4, %xmm2
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
; SSE41-NEXT:    por %xmm0, %xmm2
; SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [8192,24640,41088,57536,57600,41152,24704,8256]
; SSE41-NEXT:    pblendvb %xmm0, %xmm2, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm2
; SSE41-NEXT:    psrlw $6, %xmm2
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psllw $2, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm2, %xmm3
; SSE41-NEXT:    paddb %xmm0, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm2
; SSE41-NEXT:    paddb %xmm1, %xmm2
; SSE41-NEXT:    movdqa %xmm1, %xmm3
; SSE41-NEXT:    psrlw $7, %xmm3
; SSE41-NEXT:    pand {{.*}}(%rip), %xmm3
; SSE41-NEXT:    por %xmm2, %xmm3
; SSE41-NEXT:    paddb %xmm0, %xmm0
; SSE41-NEXT:    pblendvb %xmm0, %xmm3, %xmm1
; SSE41-NEXT:    movdqa %xmm1, %xmm0
; SSE41-NEXT:    retq
;
; AVX-LABEL: constant_rotate_v16i8:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX-NEXT:    vpsllw $4, %xmm0, %xmm2
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX-NEXT:    vmovdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,57600,41152,24704,8256]
; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpsrlw $6, %xmm0, %xmm1
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpaddb %xmm0, %xmm0, %xmm1
; AVX-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX-NEXT:    vpor %xmm3, %xmm1, %xmm1
; AVX-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: constant_rotate_v16i8:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512F-NEXT:    vpsllw $4, %xmm0, %xmm2
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512F-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX512F-NEXT:    vmovdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,57600,41152,24704,8256]
; AVX512F-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vpsrlw $6, %xmm0, %xmm1
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512F-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX512F-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX512F-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vpaddb %xmm0, %xmm0, %xmm1
; AVX512F-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512F-NEXT:    vpor %xmm3, %xmm1, %xmm1
; AVX512F-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX512F-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: constant_rotate_v16i8:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT:    vpsllw $4, %xmm0, %xmm2
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
; AVX512VL-NEXT:    vpor %xmm1, %xmm2, %xmm1
; AVX512VL-NEXT:    vmovdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,57600,41152,24704,8256]
; AVX512VL-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    vpsrlw $6, %xmm0, %xmm1
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512VL-NEXT:    vpsllw $2, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm1, %xmm3, %xmm1
; AVX512VL-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX512VL-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    vpaddb %xmm0, %xmm0, %xmm1
; AVX512VL-NEXT:    vpsrlw $7, %xmm0, %xmm3
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm3, %xmm3
; AVX512VL-NEXT:    vpor %xmm3, %xmm1, %xmm1
; AVX512VL-NEXT:    vpaddb %xmm2, %xmm2, %xmm2
; AVX512VL-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: constant_rotate_v16i8:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm1 = [8,7,6,5,4,3,2,1,0,1,2,3,4,5,6,7]
; 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:    vpsrlvw %zmm1, %zmm0, %zmm1
; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,1,2,3,4,5,6,7,8,7,6,5,4,3,2,1]
; AVX512BW-NEXT:    vpsllvw %zmm2, %zmm0, %zmm0
; AVX512BW-NEXT:    vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: constant_rotate_v16i8:
; AVX512VLBW:       # %bb.0:
; 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:    vpsrlvw {{.*}}(%rip), %ymm0, %ymm1
; AVX512VLBW-NEXT:    vpmovwb %ymm1, %xmm1
; AVX512VLBW-NEXT:    vpsllvw {{.*}}(%rip), %ymm0, %ymm0
; AVX512VLBW-NEXT:    vpmovwb %ymm0, %xmm0
; AVX512VLBW-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512VLBW-NEXT:    vzeroupper
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: constant_rotate_v16i8:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotb {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: constant_rotate_v16i8:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    movdqa {{.*#+}} xmm2 = [8192,24640,41088,57536,57600,41152,24704,8256]
; X32-SSE-NEXT:    pxor %xmm0, %xmm0
; X32-SSE-NEXT:    pxor %xmm3, %xmm3
; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm3
; X32-SSE-NEXT:    movdqa %xmm1, %xmm4
; X32-SSE-NEXT:    psrlw $4, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    movdqa %xmm1, %xmm5
; X32-SSE-NEXT:    psllw $4, %xmm5
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm5
; X32-SSE-NEXT:    por %xmm4, %xmm5
; X32-SSE-NEXT:    pand %xmm3, %xmm5
; X32-SSE-NEXT:    pandn %xmm1, %xmm3
; X32-SSE-NEXT:    por %xmm5, %xmm3
; X32-SSE-NEXT:    movdqa %xmm3, %xmm1
; X32-SSE-NEXT:    psrlw $6, %xmm1
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm1
; X32-SSE-NEXT:    movdqa %xmm3, %xmm4
; X32-SSE-NEXT:    psllw $2, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    por %xmm1, %xmm4
; X32-SSE-NEXT:    paddb %xmm2, %xmm2
; X32-SSE-NEXT:    pxor %xmm1, %xmm1
; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm1
; X32-SSE-NEXT:    pand %xmm1, %xmm4
; X32-SSE-NEXT:    pandn %xmm3, %xmm1
; X32-SSE-NEXT:    por %xmm4, %xmm1
; X32-SSE-NEXT:    movdqa %xmm1, %xmm3
; X32-SSE-NEXT:    paddb %xmm1, %xmm3
; X32-SSE-NEXT:    movdqa %xmm1, %xmm4
; X32-SSE-NEXT:    psrlw $7, %xmm4
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm4
; X32-SSE-NEXT:    por %xmm3, %xmm4
; X32-SSE-NEXT:    paddb %xmm2, %xmm2
; X32-SSE-NEXT:    pcmpgtb %xmm2, %xmm0
; X32-SSE-NEXT:    pand %xmm0, %xmm4
; X32-SSE-NEXT:    pandn %xmm1, %xmm0
; X32-SSE-NEXT:    por %xmm4, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <16 x i8> %a, <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>
  %lshr = lshr <16 x i8> %a, <i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>
  %or = or <16 x i8> %shl, %lshr
  ret <16 x i8> %or
}

;
; Uniform Constant Rotates
;

define <2 x i64> @splatconstant_rotate_v2i64(<2 x i64> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_v2i64:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psllq $14, %xmm1
; SSE-NEXT:    psrlq $50, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_v2i64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllq $14, %xmm0, %xmm1
; AVX-NEXT:    vpsrlq $50, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm0, %xmm1, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: splatconstant_rotate_v2i64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprolq $14, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatconstant_rotate_v2i64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolq $14, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatconstant_rotate_v2i64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprolq $14, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatconstant_rotate_v2i64:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolq $14, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_v2i64:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotq $14, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_v2i64:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psllq $14, %xmm1
; X32-SSE-NEXT:    psrlq $50, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <2 x i64> %a, <i64 14, i64 14>
  %lshr = lshr <2 x i64> %a, <i64 50, i64 50>
  %or = or <2 x i64> %shl, %lshr
  ret <2 x i64> %or
}

define <4 x i32> @splatconstant_rotate_v4i32(<4 x i32> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_v4i32:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrld $28, %xmm1
; SSE-NEXT:    pslld $4, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_v4i32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrld $28, %xmm0, %xmm1
; AVX-NEXT:    vpslld $4, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: splatconstant_rotate_v4i32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprold $4, %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatconstant_rotate_v4i32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprold $4, %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatconstant_rotate_v4i32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprold $4, %zmm0, %zmm0
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatconstant_rotate_v4i32:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprold $4, %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_v4i32:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotd $4, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_v4i32:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrld $28, %xmm1
; X32-SSE-NEXT:    pslld $4, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <4 x i32> %a, <i32 4, i32 4, i32 4, i32 4>
  %lshr = lshr <4 x i32> %a, <i32 28, i32 28, i32 28, i32 28>
  %or = or <4 x i32> %shl, %lshr
  ret <4 x i32> %or
}

define <8 x i16> @splatconstant_rotate_v8i16(<8 x i16> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_v8i16:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrlw $9, %xmm1
; SSE-NEXT:    psllw $7, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_v8i16:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $9, %xmm0, %xmm1
; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512-LABEL: splatconstant_rotate_v8i16:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpsrlw $9, %xmm0, %xmm1
; AVX512-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_v8i16:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotw $7, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_v8i16:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlw $9, %xmm1
; X32-SSE-NEXT:    psllw $7, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <8 x i16> %a, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>
  %lshr = lshr <8 x i16> %a, <i16 9, i16 9, i16 9, i16 9, i16 9, i16 9, i16 9, i16 9>
  %or = or <8 x i16> %shl, %lshr
  ret <8 x i16> %or
}

define <16 x i8> @splatconstant_rotate_v16i8(<16 x i8> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_v16i8:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrlw $4, %xmm1
; SSE-NEXT:    pand {{.*}}(%rip), %xmm1
; SSE-NEXT:    psllw $4, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_v16i8:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX-NEXT:    vpsllw $4, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512-LABEL: splatconstant_rotate_v16i8:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512-NEXT:    vpsllw $4, %xmm0, %xmm0
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_v16i8:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotb $4, %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_v16i8:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlw $4, %xmm1
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm1
; X32-SSE-NEXT:    psllw $4, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <16 x i8> %a, <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>
  %lshr = lshr <16 x i8> %a, <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>
  %or = or <16 x i8> %shl, %lshr
  ret <16 x i8> %or
}

;
; Masked Uniform Constant Rotates
;

define <2 x i64> @splatconstant_rotate_mask_v2i64(<2 x i64> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_mask_v2i64:
; SSE:       # %bb.0:
; SSE-NEXT:    psrlq $49, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_mask_v2i64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlq $49, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: splatconstant_rotate_mask_v2i64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprolq $15, %zmm0, %zmm0
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatconstant_rotate_mask_v2i64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprolq $15, %xmm0, %xmm0
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatconstant_rotate_mask_v2i64:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprolq $15, %zmm0, %zmm0
; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatconstant_rotate_mask_v2i64:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprolq $15, %xmm0, %xmm0
; AVX512VLBW-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_mask_v2i64:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotq $15, %xmm0, %xmm0
; XOP-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_mask_v2i64:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    psrlq $49, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <2 x i64> %a, <i64 15, i64 15>
  %lshr = lshr <2 x i64> %a, <i64 49, i64 49>
  %rmask = and <2 x i64> %lshr, <i64 255, i64 127>
  %lmask = and <2 x i64> %shl, <i64 65, i64 33>
  %or = or <2 x i64> %lmask, %rmask
  ret <2 x i64> %or
}

define <4 x i32> @splatconstant_rotate_mask_v4i32(<4 x i32> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_mask_v4i32:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrld $28, %xmm1
; SSE-NEXT:    pslld $4, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_mask_v4i32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrld $28, %xmm0, %xmm1
; AVX-NEXT:    vpslld $4, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512F-LABEL: splatconstant_rotate_mask_v4i32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512F-NEXT:    vprold $4, %zmm0, %zmm0
; AVX512F-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    retq
;
; AVX512VL-LABEL: splatconstant_rotate_mask_v4i32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vprold $4, %xmm0, %xmm0
; AVX512VL-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-NEXT:    retq
;
; AVX512BW-LABEL: splatconstant_rotate_mask_v4i32:
; AVX512BW:       # %bb.0:
; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512BW-NEXT:    vprold $4, %zmm0, %zmm0
; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512BW-NEXT:    vzeroupper
; AVX512BW-NEXT:    retq
;
; AVX512VLBW-LABEL: splatconstant_rotate_mask_v4i32:
; AVX512VLBW:       # %bb.0:
; AVX512VLBW-NEXT:    vprold $4, %xmm0, %xmm0
; AVX512VLBW-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512VLBW-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_mask_v4i32:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotd $4, %xmm0, %xmm0
; XOP-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_mask_v4i32:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrld $28, %xmm1
; X32-SSE-NEXT:    pslld $4, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <4 x i32> %a, <i32 4, i32 4, i32 4, i32 4>
  %lshr = lshr <4 x i32> %a, <i32 28, i32 28, i32 28, i32 28>
  %rmask = and <4 x i32> %lshr, <i32 127, i32 255, i32 511, i32 1023>
  %lmask = and <4 x i32> %shl, <i32 1023, i32 511, i32 255, i32 127>
  %or = or <4 x i32> %lmask, %rmask
  ret <4 x i32> %or
}

define <8 x i16> @splatconstant_rotate_mask_v8i16(<8 x i16> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_mask_v8i16:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrlw $11, %xmm1
; SSE-NEXT:    psllw $5, %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_mask_v8i16:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $11, %xmm0, %xmm1
; AVX-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512-LABEL: splatconstant_rotate_mask_v8i16:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpsrlw $11, %xmm0, %xmm1
; AVX512-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_mask_v8i16:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotw $5, %xmm0, %xmm0
; XOP-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_mask_v8i16:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlw $11, %xmm1
; X32-SSE-NEXT:    psllw $5, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <8 x i16> %a, <i16 5, i16 5, i16 5, i16 5, i16 5, i16 5, i16 5, i16 5>
  %lshr = lshr <8 x i16> %a, <i16 11, i16 11, i16 11, i16 11, i16 11, i16 11, i16 11, i16 11>
  %rmask = and <8 x i16> %lshr, <i16 55, i16 55, i16 55, i16 55, i16 55, i16 55, i16 55, i16 55>
  %lmask = and <8 x i16> %shl, <i16 33, i16 33, i16 33, i16 33, i16 33, i16 33, i16 33, i16 33>
  %or = or <8 x i16> %lmask, %rmask
  ret <8 x i16> %or
}

define <16 x i8> @splatconstant_rotate_mask_v16i8(<16 x i8> %a) nounwind {
; SSE-LABEL: splatconstant_rotate_mask_v16i8:
; SSE:       # %bb.0:
; SSE-NEXT:    movdqa %xmm0, %xmm1
; SSE-NEXT:    psrlw $4, %xmm1
; SSE-NEXT:    pand {{.*}}(%rip), %xmm1
; SSE-NEXT:    psllw $4, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    por %xmm1, %xmm0
; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: splatconstant_rotate_mask_v16i8:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX-NEXT:    vpsllw $4, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT:    retq
;
; AVX512-LABEL: splatconstant_rotate_mask_v16i8:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpsrlw $4, %xmm0, %xmm1
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
; AVX512-NEXT:    vpsllw $4, %xmm0, %xmm0
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT:    retq
;
; XOP-LABEL: splatconstant_rotate_mask_v16i8:
; XOP:       # %bb.0:
; XOP-NEXT:    vprotb $4, %xmm0, %xmm0
; XOP-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT:    retq
;
; X32-SSE-LABEL: splatconstant_rotate_mask_v16i8:
; X32-SSE:       # %bb.0:
; X32-SSE-NEXT:    movdqa %xmm0, %xmm1
; X32-SSE-NEXT:    psrlw $4, %xmm1
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm1
; X32-SSE-NEXT:    psllw $4, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    por %xmm1, %xmm0
; X32-SSE-NEXT:    pand {{\.LCPI.*}}, %xmm0
; X32-SSE-NEXT:    retl
  %shl = shl <16 x i8> %a, <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>
  %lshr = lshr <16 x i8> %a, <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>
  %rmask = and <16 x i8> %lshr, <i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55, i8 55>
  %lmask = and <16 x i8> %shl, <i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33, i8 33>
  %or = or <16 x i8> %lmask, %rmask
  ret <16 x i8> %or
}