mirror of
https://github.com/RPCS3/llvm.git
synced 2026-07-25 04:35:32 -04:00
Simon Pilgrim
d005962cad
Similar to the existing code to lower to PACKSS, we can use PACKUS if the input vector's leading zero bits extend all the way to the packed/truncated value. We have to account for pre-SSE41 targets not supporting PACKUSDW git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@317128 91177308-0d34-0410-b5e6-96231b3b80d8
50 lines
2.3 KiB
LLVM
50 lines
2.3 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512
|
|
|
|
; This matter of this test is ensuring that vpackus* is not used for umin+trunc combination, since vpackus* input is a signed number.
|
|
|
|
define <16 x i8> @usat_trunc_wb_256(<16 x i16> %i) {
|
|
; AVX-LABEL: usat_trunc_wb_256:
|
|
; AVX: # BB#0:
|
|
; AVX-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
|
|
; AVX-NEXT: vpminuw %xmm2, %xmm1, %xmm1
|
|
; AVX-NEXT: vpminuw %xmm2, %xmm0, %xmm0
|
|
; AVX-NEXT: vpackuswb %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vzeroupper
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: usat_trunc_wb_256:
|
|
; AVX512: # BB#0:
|
|
; AVX512-NEXT: vpmovuswb %ymm0, %xmm0
|
|
; AVX512-NEXT: vzeroupper
|
|
; AVX512-NEXT: retq
|
|
%x3 = icmp ult <16 x i16> %i, <i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255>
|
|
%x5 = select <16 x i1> %x3, <16 x i16> %i, <16 x i16> <i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255, i16 255>
|
|
%x6 = trunc <16 x i16> %x5 to <16 x i8>
|
|
ret <16 x i8> %x6
|
|
}
|
|
|
|
define <8 x i16> @usat_trunc_dw_256(<8 x i32> %i) {
|
|
; AVX-LABEL: usat_trunc_dw_256:
|
|
; AVX: # BB#0:
|
|
; AVX-NEXT: vextractf128 $1, %ymm0, %xmm1
|
|
; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = [65535,65535,65535,65535]
|
|
; AVX-NEXT: vpminud %xmm2, %xmm1, %xmm1
|
|
; AVX-NEXT: vpminud %xmm2, %xmm0, %xmm0
|
|
; AVX-NEXT: vpackusdw %xmm1, %xmm0, %xmm0
|
|
; AVX-NEXT: vzeroupper
|
|
; AVX-NEXT: retq
|
|
;
|
|
; AVX512-LABEL: usat_trunc_dw_256:
|
|
; AVX512: # BB#0:
|
|
; AVX512-NEXT: vpmovusdw %ymm0, %xmm0
|
|
; AVX512-NEXT: vzeroupper
|
|
; AVX512-NEXT: retq
|
|
%x3 = icmp ult <8 x i32> %i, <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
|
|
%x5 = select <8 x i1> %x3, <8 x i32> %i, <8 x i32> <i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535, i32 65535>
|
|
%x6 = trunc <8 x i32> %x5 to <8 x i16>
|
|
ret <8 x i16> %x6
|
|
}
|