derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
181 lines
6.8 KiB
YAML
181 lines
6.8 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
|
|
# RUN: llc -mtriple=aarch64-linux-gnu -run-pass=aarch64-condopt -verify-machineinstrs %s -o - | FileCheck %s
|
|
# RUN: llc -mtriple=aarch64-linux-gnu -passes=aarch64-condopt %s -o - | FileCheck %s
|
|
|
|
---
|
|
# CSEL pair — same direction (GT/GT, diff 1). The first CMP is adjusted from
|
|
# #9 to #10 and its CSEL condition relaxed GT -> GE, matching the second pair
|
|
# so CSE can eliminate the duplicate CMP.
|
|
#
|
|
# Input: CMP #9 GT; CMP #10 GT
|
|
# Output: CMP #10 GE; CMP #10 GT
|
|
name: csel_same_direction
|
|
tracksRegLiveness: true
|
|
body: |
|
|
bb.0:
|
|
liveins: $w0, $w1, $w2, $x3, $x4
|
|
|
|
; CHECK-LABEL: name: csel_same_direction
|
|
; CHECK: liveins: $w0, $w1, $w2, $x3, $x4
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr32common = COPY $w0
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr32 = COPY $w1
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr32 = COPY $w2
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:gpr64common = COPY $x3
|
|
; CHECK-NEXT: [[COPY4:%[0-9]+]]:gpr64common = COPY $x4
|
|
; CHECK-NEXT: [[SUBSWri:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSELWr:%[0-9]+]]:gpr32 = CSELWr [[COPY1]], [[COPY2]], 10, implicit $nzcv
|
|
; CHECK-NEXT: [[SUBSWri1:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSELWr1:%[0-9]+]]:gpr32 = CSELWr [[COPY1]], [[COPY2]], 12, implicit $nzcv
|
|
; CHECK-NEXT: STRWui killed [[CSELWr]], [[COPY3]], 0
|
|
; CHECK-NEXT: STRWui killed [[CSELWr1]], [[COPY4]], 0
|
|
; CHECK-NEXT: RET_ReallyLR
|
|
%0:gpr32common = COPY $w0
|
|
%1:gpr32 = COPY $w1
|
|
%2:gpr32 = COPY $w2
|
|
%3:gpr64common = COPY $x3
|
|
%4:gpr64common = COPY $x4
|
|
|
|
%5:gpr32 = SUBSWri %0, 9, 0, implicit-def $nzcv
|
|
%6:gpr32 = CSELWr %1, %2, 12, implicit $nzcv
|
|
|
|
%7:gpr32 = SUBSWri %0, 10, 0, implicit-def $nzcv
|
|
%8:gpr32 = CSELWr %1, %2, 12, implicit $nzcv
|
|
|
|
STRWui killed %6, %3, 0
|
|
STRWui killed %8, %4, 0
|
|
RET_ReallyLR
|
|
|
|
...
|
|
---
|
|
# CSINV pair — opposite direction (GT/LT, diff 2). Both CMPs are adjusted to
|
|
# #10, conditions relaxed GT -> GE and LT -> LE.
|
|
#
|
|
# Input: CMP #9 GT; CMP #11 LT
|
|
# Output: CMP #10 GE; CMP #10 LE
|
|
name: csinv_opposite_direction
|
|
tracksRegLiveness: true
|
|
body: |
|
|
bb.0:
|
|
liveins: $w0, $w1, $w2, $x3, $x4
|
|
|
|
; CHECK-LABEL: name: csinv_opposite_direction
|
|
; CHECK: liveins: $w0, $w1, $w2, $x3, $x4
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr32common = COPY $w0
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr32 = COPY $w1
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr32 = COPY $w2
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:gpr64common = COPY $x3
|
|
; CHECK-NEXT: [[COPY4:%[0-9]+]]:gpr64common = COPY $x4
|
|
; CHECK-NEXT: [[SUBSWri:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSINVWr:%[0-9]+]]:gpr32 = CSINVWr [[COPY1]], [[COPY2]], 10, implicit $nzcv
|
|
; CHECK-NEXT: [[SUBSWri1:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSINVWr1:%[0-9]+]]:gpr32 = CSINVWr [[COPY1]], [[COPY2]], 13, implicit $nzcv
|
|
; CHECK-NEXT: STRWui killed [[CSINVWr]], [[COPY3]], 0
|
|
; CHECK-NEXT: STRWui killed [[CSINVWr1]], [[COPY4]], 0
|
|
; CHECK-NEXT: RET_ReallyLR
|
|
%0:gpr32common = COPY $w0
|
|
%1:gpr32 = COPY $w1
|
|
%2:gpr32 = COPY $w2
|
|
%3:gpr64common = COPY $x3
|
|
%4:gpr64common = COPY $x4
|
|
|
|
%5:gpr32 = SUBSWri %0, 9, 0, implicit-def $nzcv
|
|
%6:gpr32 = CSINVWr %1, %2, 12, implicit $nzcv
|
|
|
|
%7:gpr32 = SUBSWri %0, 11, 0, implicit-def $nzcv
|
|
%8:gpr32 = CSINVWr %1, %2, 11, implicit $nzcv
|
|
|
|
STRWui killed %6, %3, 0
|
|
STRWui killed %8, %4, 0
|
|
RET_ReallyLR
|
|
|
|
...
|
|
---
|
|
# CSNEG pair — same direction (GT/GT, diff 1).
|
|
#
|
|
# Input: CMP #9 GT; CMP #10 GT
|
|
# Output: CMP #10 GE; CMP #10 GT
|
|
name: csneg_same_direction
|
|
tracksRegLiveness: true
|
|
body: |
|
|
bb.0:
|
|
liveins: $w0, $w1, $w2, $x3, $x4
|
|
|
|
; CHECK-LABEL: name: csneg_same_direction
|
|
; CHECK: liveins: $w0, $w1, $w2, $x3, $x4
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr32common = COPY $w0
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr32 = COPY $w1
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr32 = COPY $w2
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:gpr64common = COPY $x3
|
|
; CHECK-NEXT: [[COPY4:%[0-9]+]]:gpr64common = COPY $x4
|
|
; CHECK-NEXT: [[SUBSWri:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSNEGWr:%[0-9]+]]:gpr32 = CSNEGWr [[COPY1]], [[COPY2]], 10, implicit $nzcv
|
|
; CHECK-NEXT: [[SUBSWri1:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSNEGWr1:%[0-9]+]]:gpr32 = CSNEGWr [[COPY1]], [[COPY2]], 12, implicit $nzcv
|
|
; CHECK-NEXT: STRWui killed [[CSNEGWr]], [[COPY3]], 0
|
|
; CHECK-NEXT: STRWui killed [[CSNEGWr1]], [[COPY4]], 0
|
|
; CHECK-NEXT: RET_ReallyLR
|
|
%0:gpr32common = COPY $w0
|
|
%1:gpr32 = COPY $w1
|
|
%2:gpr32 = COPY $w2
|
|
%3:gpr64common = COPY $x3
|
|
%4:gpr64common = COPY $x4
|
|
|
|
%5:gpr32 = SUBSWri %0, 9, 0, implicit-def $nzcv
|
|
%6:gpr32 = CSNEGWr %1, %2, 12, implicit $nzcv
|
|
|
|
%7:gpr32 = SUBSWri %0, 10, 0, implicit-def $nzcv
|
|
%8:gpr32 = CSNEGWr %1, %2, 12, implicit $nzcv
|
|
|
|
STRWui killed %6, %3, 0
|
|
STRWui killed %8, %4, 0
|
|
RET_ReallyLR
|
|
|
|
...
|
|
---
|
|
# Mixed: CSINC and CSEL in the same block. Both pairs should be optimized
|
|
# independently regardless of instruction type.
|
|
#
|
|
# Input: CMP #9 GT (CSINC); CMP #10 GT (CSEL)
|
|
# Output: CMP #10 GE (CSINC); CMP #10 GT (CSEL)
|
|
name: mixed_csinc_csel
|
|
tracksRegLiveness: true
|
|
body: |
|
|
bb.0:
|
|
liveins: $w0, $w1, $w2, $x3, $x4
|
|
|
|
; CHECK-LABEL: name: mixed_csinc_csel
|
|
; CHECK: liveins: $w0, $w1, $w2, $x3, $x4
|
|
; CHECK-NEXT: {{ $}}
|
|
; CHECK-NEXT: [[COPY:%[0-9]+]]:gpr32common = COPY $w0
|
|
; CHECK-NEXT: [[COPY1:%[0-9]+]]:gpr32 = COPY $w1
|
|
; CHECK-NEXT: [[COPY2:%[0-9]+]]:gpr32 = COPY $w2
|
|
; CHECK-NEXT: [[COPY3:%[0-9]+]]:gpr64common = COPY $x3
|
|
; CHECK-NEXT: [[COPY4:%[0-9]+]]:gpr64common = COPY $x4
|
|
; CHECK-NEXT: [[SUBSWri:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSINCWr:%[0-9]+]]:gpr32 = CSINCWr [[COPY1]], [[COPY2]], 10, implicit $nzcv
|
|
; CHECK-NEXT: [[SUBSWri1:%[0-9]+]]:gpr32 = SUBSWri [[COPY]], 10, 0, implicit-def $nzcv
|
|
; CHECK-NEXT: [[CSELWr:%[0-9]+]]:gpr32 = CSELWr [[COPY1]], [[COPY2]], 12, implicit $nzcv
|
|
; CHECK-NEXT: STRWui killed [[CSINCWr]], [[COPY3]], 0
|
|
; CHECK-NEXT: STRWui killed [[CSELWr]], [[COPY4]], 0
|
|
; CHECK-NEXT: RET_ReallyLR
|
|
%0:gpr32common = COPY $w0
|
|
%1:gpr32 = COPY $w1
|
|
%2:gpr32 = COPY $w2
|
|
%3:gpr64common = COPY $x3
|
|
%4:gpr64common = COPY $x4
|
|
|
|
%5:gpr32 = SUBSWri %0, 9, 0, implicit-def $nzcv
|
|
%6:gpr32 = CSINCWr %1, %2, 12, implicit $nzcv
|
|
|
|
%7:gpr32 = SUBSWri %0, 10, 0, implicit-def $nzcv
|
|
%8:gpr32 = CSELWr %1, %2, 12, implicit $nzcv
|
|
|
|
STRWui killed %6, %3, 0
|
|
STRWui killed %8, %4, 0
|
|
RET_ReallyLR
|
|
|
|
...
|