mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-21 20:12:56 +02:00
24ce82aeec
That's not the case for VPERMV/VPERMV3, which cover all possible combinations (the C intrinsics use a different order; the AVX vs AVX512 intrinsics are different still). Since: r246981 AVX-512: Lowering for 512-bit vector shuffles. VPERMV is recognized in getTargetShuffleMask. This breaks assumptions in most callers, as they expect the non-mask operands to start at index 0. VPERMV has the mask as operand #0; VPERMV3 has it in the middle. Instead of the faulty assumption, have getTargetShuffleMask return its operands as well. One alternative we considered was to change the operand order of VPERMV, but we agreed to stick to the instruction order, as there are more AVX512 weirdness to cover (vpermt2/vpermi2 in particular). Differential Revision: http://reviews.llvm.org/D17041 llvm-svn: 262627
28 lines
1.8 KiB
LLVM
28 lines
1.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx2 | FileCheck %s
|
|
|
|
declare <8 x i32> @llvm.x86.avx2.permd(<8 x i32>, <8 x i32>)
|
|
declare <8 x float> @llvm.x86.avx2.permps(<8 x float>, <8 x i32>)
|
|
|
|
define <32 x i8> @combine_pshufb_vpermd(<8 x i32> %a) {
|
|
; CHECK-LABEL: combine_pshufb_vpermd:
|
|
; CHECK: # BB#0:
|
|
; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18]
|
|
; CHECK-NEXT: retq
|
|
%tmp0 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 4>)
|
|
%tmp1 = bitcast <8 x i32> %tmp0 to <32 x i8>
|
|
%tmp2 = shufflevector <32 x i8> %tmp1, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 30>
|
|
ret <32 x i8> %tmp2
|
|
}
|
|
|
|
define <32 x i8> @combine_pshufb_vpermps(<8 x float> %a) {
|
|
; CHECK-LABEL: combine_pshufb_vpermps:
|
|
; CHECK: # BB#0:
|
|
; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18]
|
|
; CHECK-NEXT: retq
|
|
%tmp0 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 4>)
|
|
%tmp1 = bitcast <8 x float> %tmp0 to <32 x i8>
|
|
%tmp2 = shufflevector <32 x i8> %tmp1, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 30>
|
|
ret <32 x i8> %tmp2
|
|
}
|