forked from OSchip/llvm-project
53 lines
1.9 KiB
Plaintext
53 lines
1.9 KiB
Plaintext
|
# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx -global-isel -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=AVX
|
||
|
# RUN: llc -mtriple=x86_64-linux-gnu -mattr=+avx512f,+avx512vl -global-isel -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=AVX512VL
|
||
|
--- |
|
||
|
define void @test_merge() {
|
||
|
ret void
|
||
|
}
|
||
|
...
|
||
|
---
|
||
|
name: test_merge
|
||
|
# AVX-LABEL: name: test_merge
|
||
|
#
|
||
|
# AVX512VL-LABEL: name: test_merge
|
||
|
alignment: 4
|
||
|
legalized: true
|
||
|
regBankSelected: true
|
||
|
# AVX: registers:
|
||
|
# AVX-NEXT: - { id: 0, class: vr128, preferred-register: '' }
|
||
|
# AVX-NEXT: - { id: 1, class: vr256, preferred-register: '' }
|
||
|
# AVX-NEXT: - { id: 2, class: vr256, preferred-register: '' }
|
||
|
# AVX-NEXT: - { id: 3, class: vr256, preferred-register: '' }
|
||
|
#
|
||
|
# AVX512VL: registers:
|
||
|
# AVX512VL-NEXT: - { id: 0, class: vr128x, preferred-register: '' }
|
||
|
# AVX512VL-NEXT: - { id: 1, class: vr256x, preferred-register: '' }
|
||
|
# AVX512VL-NEXT: - { id: 2, class: vr256x, preferred-register: '' }
|
||
|
# AVX512VL-NEXT: - { id: 3, class: vr256x, preferred-register: '' }
|
||
|
registers:
|
||
|
- { id: 0, class: vecr }
|
||
|
- { id: 1, class: vecr }
|
||
|
# AVX: %0 = IMPLICIT_DEF
|
||
|
# AVX-NEXT: undef %2.sub_xmm = COPY %0
|
||
|
# AVX-NEXT: %3 = VINSERTF128rr %2, %0, 1
|
||
|
# AVX-NEXT: %1 = COPY %3
|
||
|
# AVX-NEXT: %ymm0 = COPY %1
|
||
|
# AVX-NEXT: RET 0, implicit %ymm0
|
||
|
#
|
||
|
# AVX512VL: %0 = IMPLICIT_DEF
|
||
|
# AVX512VL-NEXT: undef %2.sub_xmm = COPY %0
|
||
|
# AVX512VL-NEXT: %3 = VINSERTF32x4Z256rr %2, %0, 1
|
||
|
# AVX512VL-NEXT: %1 = COPY %3
|
||
|
# AVX512VL-NEXT: %ymm0 = COPY %1
|
||
|
# AVX512VL-NEXT: RET 0, implicit %ymm0
|
||
|
body: |
|
||
|
bb.1 (%ir-block.0):
|
||
|
|
||
|
%0(<4 x s32>) = IMPLICIT_DEF
|
||
|
%1(<8 x s32>) = G_MERGE_VALUES %0(<4 x s32>), %0(<4 x s32>)
|
||
|
%ymm0 = COPY %1(<8 x s32>)
|
||
|
RET 0, implicit %ymm0
|
||
|
|
||
|
...
|
||
|
|