maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[X86] Add AVX512 tests for vector rotations

This commit is contained in:
Simon Pilgrim 2020-01-23 17:17:24 +00:00
parent e25eee4db7
commit 98e37af272
1 changed files with 42 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

61
llvm/test/CodeGen/X86/rotate_vec.ll
View File

@ -1,11 +1,17 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=bdver4 | FileCheck %s
; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=bdver4 | FileCheck %s --check-prefixes=CHECK,XOP
; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=skylake-avx512 | FileCheck %s --check-prefixes=CHECK,AVX512
define <4 x i32> @rot_v4i32_splat(<4 x i32> %x) {
; CHECK-LABEL: rot_v4i32_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vprotd $31, %xmm0, %xmm0
; CHECK-NEXT: retq
; XOP-LABEL: rot_v4i32_splat:
; XOP: # %bb.0:
; XOP-NEXT: vprotd $31, %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: rot_v4i32_splat:
; AVX512: # %bb.0:
; AVX512-NEXT: vprold $31, %xmm0, %xmm0
; AVX512-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 1, i32 1, i32 1, i32 1>
%2 = shl <4 x i32> %x, <i32 31, i32 31, i32 31, i32 31>
%3 = or <4 x i32> %1, %2
@ -13,10 +19,15 @@ define <4 x i32> @rot_v4i32_splat(<4 x i32> %x) {
}
define <4 x i32> @rot_v4i32_non_splat(<4 x i32> %x) {
; CHECK-LABEL: rot_v4i32_non_splat:
; CHECK: # %bb.0:
; CHECK-NEXT: vprotd {{.*}}(%rip), %xmm0, %xmm0
; CHECK-NEXT: retq
; XOP-LABEL: rot_v4i32_non_splat:
; XOP: # %bb.0:
; XOP-NEXT: vprotd {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: rot_v4i32_non_splat:
; AVX512: # %bb.0:
; AVX512-NEXT: vprolvd {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 1, i32 2, i32 3, i32 4>
%2 = shl <4 x i32> %x, <i32 31, i32 30, i32 29, i32 28>
%3 = or <4 x i32> %1, %2
@ -24,11 +35,17 @@ define <4 x i32> @rot_v4i32_non_splat(<4 x i32> %x) {
}
define <4 x i32> @rot_v4i32_splat_2masks(<4 x i32> %x) {
; CHECK-LABEL: rot_v4i32_splat_2masks:
; CHECK: # %bb.0:
; CHECK-NEXT: vprotd $31, %xmm0, %xmm0
; CHECK-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; CHECK-NEXT: retq
; XOP-LABEL: rot_v4i32_splat_2masks:
; XOP: # %bb.0:
; XOP-NEXT: vprotd $31, %xmm0, %xmm0
; XOP-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: rot_v4i32_splat_2masks:
; AVX512: # %bb.0:
; AVX512-NEXT: vprold $31, %xmm0, %xmm0
; AVX512-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 1, i32 1, i32 1, i32 1>
%2 = and <4 x i32> %1, <i32 4294901760, i32 4294901760, i32 4294901760, i32 4294901760>
@ -39,11 +56,17 @@ define <4 x i32> @rot_v4i32_splat_2masks(<4 x i32> %x) {
}
define <4 x i32> @rot_v4i32_non_splat_2masks(<4 x i32> %x) {
; CHECK-LABEL: rot_v4i32_non_splat_2masks:
; CHECK: # %bb.0:
; CHECK-NEXT: vprotd {{.*}}(%rip), %xmm0, %xmm0
; CHECK-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; CHECK-NEXT: retq
; XOP-LABEL: rot_v4i32_non_splat_2masks:
; XOP: # %bb.0:
; XOP-NEXT: vprotd {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; XOP-NEXT: retq
;
; AVX512-LABEL: rot_v4i32_non_splat_2masks:
; AVX512: # %bb.0:
; AVX512-NEXT: vprolvd {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: vpand {{.*}}(%rip), %xmm0, %xmm0
; AVX512-NEXT: retq
%1 = lshr <4 x i32> %x, <i32 1, i32 2, i32 3, i32 4>
%2 = and <4 x i32> %1, <i32 4294901760, i32 4294901760, i32 4294901760, i32 4294901760>