Skip to content

Commit

Permalink
[X86][AVX512DQ] Added AVX512DQ to 128/256 bit vector shift tests
Browse files Browse the repository at this point in the history 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291444 91177308-0d34-0410-b5e6-96231b3b80d8
  • Loading branch information
@RKSimon
RKSimon committed Jan 9, 2017
1 parent a2d2e19 commit 077d763
Show file tree
Hide file tree
Showing 6 changed files with 215 additions and 84 deletions.
46 changes: 32 additions & 14 deletions test/CodeGen/X86/vector-shift-ashr-128.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW

;
Expand Down Expand Up @@ -321,13 +322,22 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; XOP-NEXT: vpshaw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: var_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-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: vpmovsxwd %xmm0, %ymm0
; AVX512DQ-NEXT: vpsravd %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: vpmovdw %zmm0, %ymm0
; AVX512DQ-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill>
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: var_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down Expand Up @@ -1221,13 +1231,21 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) nounwind {
; XOP-NEXT: vpshaw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: constant_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: constant_shift_v8i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpmovsxwd %xmm0, %ymm0
; AVX512DQ-NEXT: vpsravd {{.*}}(%rip), %ymm0, %ymm0
; AVX512DQ-NEXT: vpmovdw %zmm0, %ymm0
; AVX512DQ-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill>
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: constant_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down
58 changes: 44 additions & 14 deletions test/CodeGen/X86/vector-shift-ashr-256.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
;
; Variable Shifts
Expand Down Expand Up @@ -212,13 +213,27 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind {
; XOPAVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: var_shift_v16i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %YMM1<def> %YMM1<kill> %ZMM1<def>
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: var_shift_v16i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpxor %ymm2, %ymm2, %ymm2
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm3 = ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm4 = ymm2[4],ymm0[4],ymm2[5],ymm0[5],ymm2[6],ymm0[6],ymm2[7],ymm0[7],ymm2[12],ymm0[12],ymm2[13],ymm0[13],ymm2[14],ymm0[14],ymm2[15],ymm0[15]
; AVX512DQ-NEXT: vpsravd %ymm3, %ymm4, %ymm3
; AVX512DQ-NEXT: vpsrld $16, %ymm3, %ymm3
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm1 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11]
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm2[0],ymm0[0],ymm2[1],ymm0[1],ymm2[2],ymm0[2],ymm2[3],ymm0[3],ymm2[8],ymm0[8],ymm2[9],ymm0[9],ymm2[10],ymm0[10],ymm2[11],ymm0[11]
; AVX512DQ-NEXT: vpsravd %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512DQ-NEXT: vpackusdw %ymm3, %ymm0, %ymm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v16i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %YMM1<def> %YMM1<kill> %ZMM1<def>
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
%shift = ashr <16 x i16> %a, %b
ret <16 x i16> %shift
}
Expand Down Expand Up @@ -804,13 +819,28 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) nounwind {
; XOPAVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: constant_shift_v16i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: constant_shift_v16i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
; AVX512DQ-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm3 = ymm2[4],ymm1[4],ymm2[5],ymm1[5],ymm2[6],ymm1[6],ymm2[7],ymm1[7],ymm2[12],ymm1[12],ymm2[13],ymm1[13],ymm2[14],ymm1[14],ymm2[15],ymm1[15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm4 = ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15]
; AVX512DQ-NEXT: vpsravd %ymm3, %ymm4, %ymm3
; AVX512DQ-NEXT: vpsrld $16, %ymm3, %ymm3
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm2 = ymm2[0],ymm1[0],ymm2[1],ymm1[1],ymm2[2],ymm1[2],ymm2[3],ymm1[3],ymm2[8],ymm1[8],ymm2[9],ymm1[9],ymm2[10],ymm1[10],ymm2[11],ymm1[11]
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11]
; AVX512DQ-NEXT: vpsravd %ymm2, %ymm0, %ymm0
; AVX512DQ-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512DQ-NEXT: vpackusdw %ymm3, %ymm0, %ymm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v16i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
%shift = ashr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>
ret <16 x i16> %shift
}
Expand Down
46 changes: 32 additions & 14 deletions test/CodeGen/X86/vector-shift-lshr-128.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
;
; Just one 32-bit run to make sure we do reasonable things for i64 shifts.
Expand Down Expand Up @@ -290,13 +291,22 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; XOP-NEXT: vpshlw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: var_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-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: vpsrlvd %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: vpmovdw %zmm0, %ymm0
; AVX512DQ-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill>
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: var_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down Expand Up @@ -955,13 +965,21 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) nounwind {
; XOP-NEXT: vpshlw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: constant_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: constant_shift_v8i16:
; AVX512DQ: # BB#0:
; 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: vpsrlvd {{.*}}(%rip), %ymm0, %ymm0
; AVX512DQ-NEXT: vpmovdw %zmm0, %ymm0
; AVX512DQ-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill>
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512BW-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: constant_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down
58 changes: 44 additions & 14 deletions test/CodeGen/X86/vector-shift-lshr-256.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
;
; Variable Shifts
Expand Down Expand Up @@ -189,13 +190,27 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) nounwind {
; XOPAVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: var_shift_v16i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %YMM1<def> %YMM1<kill> %ZMM1<def>
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: var_shift_v16i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpxor %ymm2, %ymm2, %ymm2
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm3 = ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm4 = ymm2[4],ymm0[4],ymm2[5],ymm0[5],ymm2[6],ymm0[6],ymm2[7],ymm0[7],ymm2[12],ymm0[12],ymm2[13],ymm0[13],ymm2[14],ymm0[14],ymm2[15],ymm0[15]
; AVX512DQ-NEXT: vpsrlvd %ymm3, %ymm4, %ymm3
; AVX512DQ-NEXT: vpsrld $16, %ymm3, %ymm3
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm1 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11]
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm2[0],ymm0[0],ymm2[1],ymm0[1],ymm2[2],ymm0[2],ymm2[3],ymm0[3],ymm2[8],ymm0[8],ymm2[9],ymm0[9],ymm2[10],ymm0[10],ymm2[11],ymm0[11]
; AVX512DQ-NEXT: vpsrlvd %ymm1, %ymm0, %ymm0
; AVX512DQ-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512DQ-NEXT: vpackusdw %ymm3, %ymm0, %ymm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v16i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %YMM1<def> %YMM1<kill> %ZMM1<def>
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
%shift = lshr <16 x i16> %a, %b
ret <16 x i16> %shift
}
Expand Down Expand Up @@ -659,13 +674,28 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) nounwind {
; XOPAVX2-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
; XOPAVX2-NEXT: retq
;
; AVX512-LABEL: constant_shift_v16i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: constant_shift_v16i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpxor %ymm1, %ymm1, %ymm1
; AVX512DQ-NEXT: vmovdqa {{.*#+}} ymm2 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm3 = ymm2[4],ymm1[4],ymm2[5],ymm1[5],ymm2[6],ymm1[6],ymm2[7],ymm1[7],ymm2[12],ymm1[12],ymm2[13],ymm1[13],ymm2[14],ymm1[14],ymm2[15],ymm1[15]
; AVX512DQ-NEXT: vpunpckhwd {{.*#+}} ymm4 = ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15]
; AVX512DQ-NEXT: vpsrlvd %ymm3, %ymm4, %ymm3
; AVX512DQ-NEXT: vpsrld $16, %ymm3, %ymm3
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm2 = ymm2[0],ymm1[0],ymm2[1],ymm1[1],ymm2[2],ymm1[2],ymm2[3],ymm1[3],ymm2[8],ymm1[8],ymm2[9],ymm1[9],ymm2[10],ymm1[10],ymm2[11],ymm1[11]
; AVX512DQ-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11]
; AVX512DQ-NEXT: vpsrlvd %ymm2, %ymm0, %ymm0
; AVX512DQ-NEXT: vpsrld $16, %ymm0, %ymm0
; AVX512DQ-NEXT: vpackusdw %ymm3, %ymm0, %ymm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v16i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15]
; AVX512BW-NEXT: vpsrlvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
%shift = lshr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>
ret <16 x i16> %shift
}
Expand Down
43 changes: 29 additions & 14 deletions test/CodeGen/X86/vector-shift-shl-128.ll
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+xop,+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=XOP --check-prefix=XOPAVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512dq | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=knl -mattr=+avx512bw | FileCheck %s --check-prefix=ALL --check-prefix=AVX512 --check-prefix=AVX512BW
;
; Just one 32-bit run to make sure we do reasonable things for i64 shifts.
Expand Down Expand Up @@ -245,13 +246,22 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) nounwind {
; XOP-NEXT: vpshlw %xmm1, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: var_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-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: %XMM0<def> %XMM0<kill> %YMM0<kill>
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: var_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def>
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: var_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down Expand Up @@ -827,13 +837,18 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) nounwind {
; XOP-NEXT: vpshlw {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: constant_shift_v8i16:
; AVX512: # BB#0:
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512-NEXT: retq
; AVX512DQ-LABEL: constant_shift_v8i16:
; AVX512DQ: # BB#0:
; AVX512DQ-NEXT: vpmullw {{.*}}(%rip), %xmm0, %xmm0
; AVX512DQ-NEXT: retq
;
; AVX512BW-LABEL: constant_shift_v8i16:
; AVX512BW: # BB#0:
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def>
; AVX512BW-NEXT: vmovdqa {{.*#+}} xmm1 = [0,1,2,3,4,5,6,7]
; AVX512BW-NEXT: vpsllvw %zmm1, %zmm0, %zmm0
; AVX512BW-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<kill>
; AVX512BW-NEXT: retq
;
; X32-SSE-LABEL: constant_shift_v8i16:
; X32-SSE: # BB#0:
Expand Down
Loading

0 comments on commit 077d763

Please sign in to comment.