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
70 lines
3.5 KiB
LLVM
70 lines
3.5 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
|
|
; Check that the default heuristic use the local cse constraints.
|
|
; RUN: opt -S -passes=reassociate %s -o - | FileCheck %s -check-prefix=LOCAL_CSE
|
|
; RUN: opt -S -passes=reassociate %s -reassociate-use-cse-local=true -o - | FileCheck %s -check-prefix=LOCAL_CSE
|
|
; RUN: opt -S -passes=reassociate %s -reassociate-use-cse-local=false -o - | FileCheck %s -check-prefix=CSE
|
|
|
|
; In the local CSE mode, we check that we don't create loop dependent
|
|
; expressions in order to expose more CSE opportunities.
|
|
; This can be seen with %inv4 and %inv5 that should stay in the entry block.
|
|
;
|
|
; For the non-local CSE mode, we form CSE-able expression regardless of where
|
|
; they would be materialized. In this case, %inv4 and %inv5 are pushed
|
|
; down the loop body in order to make loop_dependent, %inv2 appear as a
|
|
; sub-expression.
|
|
;
|
|
; Related to issue PR61458.
|
|
define void @reassoc_defeats_licm(i64 %inv1, i64 %inv2, i64 %inv3) {
|
|
; LOCAL_CSE-LABEL: define void @reassoc_defeats_licm
|
|
; LOCAL_CSE-SAME: (i64 [[INV1:%.*]], i64 [[INV2:%.*]], i64 [[INV3:%.*]]) {
|
|
; LOCAL_CSE-NEXT: bb:
|
|
; LOCAL_CSE-NEXT: [[INV4:%.*]] = add nuw nsw i64 [[INV2]], [[INV1]]
|
|
; LOCAL_CSE-NEXT: [[INV5:%.*]] = add nuw nsw i64 [[INV3]], [[INV2]]
|
|
; LOCAL_CSE-NEXT: br label [[BB214:%.*]]
|
|
; LOCAL_CSE: bb214:
|
|
; LOCAL_CSE-NEXT: [[IV1:%.*]] = phi i64 [ [[IV2:%.*]], [[BB214]] ], [ 0, [[BB:%.*]] ]
|
|
; LOCAL_CSE-NEXT: [[IV2]] = phi i64 [ [[IV2_PLUS_1:%.*]], [[BB214]] ], [ 1, [[BB]] ]
|
|
; LOCAL_CSE-NEXT: [[LOOP_DEPENDENT:%.*]] = shl nuw nsw i64 [[IV1]], 13
|
|
; LOCAL_CSE-NEXT: [[LOOP_DEPENDENT2:%.*]] = add nsw i64 [[INV4]], [[LOOP_DEPENDENT]]
|
|
; LOCAL_CSE-NEXT: call void @keep_alive(i64 [[LOOP_DEPENDENT2]])
|
|
; LOCAL_CSE-NEXT: [[LOOP_DEPENDENT3:%.*]] = add i64 [[INV5]], [[LOOP_DEPENDENT]]
|
|
; LOCAL_CSE-NEXT: call void @keep_alive(i64 [[LOOP_DEPENDENT3]])
|
|
; LOCAL_CSE-NEXT: [[IV2_PLUS_1]] = add i64 [[IV2]], 1
|
|
; LOCAL_CSE-NEXT: br label [[BB214]]
|
|
;
|
|
; CSE-LABEL: define void @reassoc_defeats_licm
|
|
; CSE-SAME: (i64 [[INV1:%.*]], i64 [[INV2:%.*]], i64 [[INV3:%.*]]) {
|
|
; CSE-NEXT: bb:
|
|
; CSE-NEXT: br label [[BB214:%.*]]
|
|
; CSE: bb214:
|
|
; CSE-NEXT: [[IV1:%.*]] = phi i64 [ [[IV2:%.*]], [[BB214]] ], [ 0, [[BB:%.*]] ]
|
|
; CSE-NEXT: [[IV2]] = phi i64 [ [[IV2_PLUS_1:%.*]], [[BB214]] ], [ 1, [[BB]] ]
|
|
; CSE-NEXT: [[LOOP_DEPENDENT:%.*]] = shl nuw nsw i64 [[IV1]], 13
|
|
; CSE-NEXT: [[INV4:%.*]] = add i64 [[LOOP_DEPENDENT]], [[INV2]]
|
|
; CSE-NEXT: [[LOOP_DEPENDENT2:%.*]] = add i64 [[INV4]], [[INV1]]
|
|
; CSE-NEXT: call void @keep_alive(i64 [[LOOP_DEPENDENT2]])
|
|
; CSE-NEXT: [[INV5:%.*]] = add i64 [[LOOP_DEPENDENT]], [[INV2]]
|
|
; CSE-NEXT: [[LOOP_DEPENDENT3:%.*]] = add i64 [[INV5]], [[INV3]]
|
|
; CSE-NEXT: call void @keep_alive(i64 [[LOOP_DEPENDENT3]])
|
|
; CSE-NEXT: [[IV2_PLUS_1]] = add i64 [[IV2]], 1
|
|
; CSE-NEXT: br label [[BB214]]
|
|
;
|
|
bb:
|
|
%inv4 = add nuw nsw i64 %inv1, %inv2
|
|
%inv5 = add nuw nsw i64 %inv2, %inv3
|
|
br label %bb214
|
|
|
|
bb214: ; preds = %bb214, %bb
|
|
%iv1 = phi i64 [ %iv2, %bb214 ], [ 0, %bb ]
|
|
%iv2 = phi i64 [ %iv2_plus_1, %bb214 ], [ 1, %bb ]
|
|
%loop_dependent = shl nuw nsw i64 %iv1, 13
|
|
%loop_dependent2 = add nsw i64 %inv4, %loop_dependent
|
|
call void @keep_alive(i64 %loop_dependent2)
|
|
%loop_dependent3 = add i64 %inv5, %loop_dependent
|
|
call void @keep_alive(i64 %loop_dependent3)
|
|
%iv2_plus_1 = add i64 %iv2, 1
|
|
br label %bb214
|
|
}
|
|
|
|
declare void @keep_alive(i64)
|