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
120 lines
6.5 KiB
LLVM
120 lines
6.5 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: opt -passes="default<O1>" -S < %s | FileCheck --check-prefixes=O1 %s
|
|
; RUN: opt -passes="default<O2>" -S < %s | FileCheck --check-prefixes=O23 %s
|
|
; RUN: opt -passes="default<O3>" -S < %s | FileCheck --check-prefixes=O23 %s
|
|
|
|
target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
|
|
target triple = "x86_64-unknown-linux-gnu"
|
|
|
|
; We should retain the TBAA on the load here, not lose it.
|
|
|
|
define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
|
|
; O1-LABEL: define void @licm(
|
|
; O1-SAME: ptr readonly align 8 captures(none) dereferenceable(8) [[_M_START_I:%.*]], i64 [[NUMELEM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
|
|
; O1-NEXT: [[ENTRY:.*:]]
|
|
; O1-NEXT: [[CMP1_NOT:%.*]] = icmp eq i64 [[NUMELEM]], 0
|
|
; O1-NEXT: br i1 [[CMP1_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_LR_PH:.*]]
|
|
; O1: [[FOR_BODY_LR_PH]]:
|
|
; O1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_START_I]], align 8, !tbaa [[ANYPTR_TBAA3:![0-9]+]]
|
|
; O1-NEXT: br label %[[FOR_BODY:.*]]
|
|
; O1: [[FOR_BODY]]:
|
|
; O1-NEXT: [[K_02:%.*]] = phi i64 [ 0, %[[FOR_BODY_LR_PH]] ], [ [[INC:%.*]], %[[FOR_BODY]] ]
|
|
; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds [8 x i8], ptr [[TMP0]], i64 [[K_02]]
|
|
; O1-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA8:![0-9]+]]
|
|
; O1-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
|
|
; O1-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
|
|
; O1-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]]
|
|
; O1: [[FOR_COND_CLEANUP]]:
|
|
; O1-NEXT: ret void
|
|
;
|
|
; O23-LABEL: define void @licm(
|
|
; O23-SAME: ptr readonly align 8 captures(none) dereferenceable(8) [[_M_START_I:%.*]], i64 [[NUMELEM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
|
|
; O23-NEXT: [[ENTRY:.*:]]
|
|
; O23-NEXT: [[CMP1_NOT:%.*]] = icmp eq i64 [[NUMELEM]], 0
|
|
; O23-NEXT: br i1 [[CMP1_NOT]], label %[[FOR_COND_CLEANUP:.*]], label %[[FOR_BODY_LR_PH:.*]]
|
|
; O23: [[FOR_BODY_LR_PH]]:
|
|
; O23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_START_I]], align 8, !tbaa [[ANYPTR_TBAA3:![0-9]+]]
|
|
; O23-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[NUMELEM]], 4
|
|
; O23-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[FOR_BODY_PREHEADER:.*]], label %[[VECTOR_PH:.*]]
|
|
; O23: [[VECTOR_PH]]:
|
|
; O23-NEXT: [[N_VEC:%.*]] = and i64 [[NUMELEM]], -4
|
|
; O23-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; O23: [[VECTOR_BODY]]:
|
|
; O23-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; O23-NEXT: [[TMP1:%.*]] = getelementptr inbounds [8 x i8], ptr [[TMP0]], i64 [[INDEX]]
|
|
; O23-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
|
|
; O23-NEXT: store <2 x double> splat (double 2.000000e+00), ptr [[TMP1]], align 8, !tbaa [[DOUBLE_TBAA8:![0-9]+]]
|
|
; O23-NEXT: store <2 x double> splat (double 2.000000e+00), ptr [[TMP2]], align 8, !tbaa [[DOUBLE_TBAA8]]
|
|
; O23-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
|
|
; O23-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
|
|
; O23-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
|
|
; O23: [[MIDDLE_BLOCK]]:
|
|
; O23-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[NUMELEM]], [[N_VEC]]
|
|
; O23-NEXT: br i1 [[CMP_N]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY_PREHEADER]]
|
|
; O23: [[FOR_BODY_PREHEADER]]:
|
|
; O23-NEXT: [[K_02_PH:%.*]] = phi i64 [ 0, %[[FOR_BODY_LR_PH]] ], [ [[N_VEC]], %[[MIDDLE_BLOCK]] ]
|
|
; O23-NEXT: br label %[[FOR_BODY:.*]]
|
|
; O23: [[FOR_BODY]]:
|
|
; O23-NEXT: [[K_02:%.*]] = phi i64 [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[K_02_PH]], %[[FOR_BODY_PREHEADER]] ]
|
|
; O23-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds [8 x i8], ptr [[TMP0]], i64 [[K_02]]
|
|
; O23-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA8]]
|
|
; O23-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
|
|
; O23-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
|
|
; O23-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
|
|
; O23: [[FOR_COND_CLEANUP]]:
|
|
; O23-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %for.cond
|
|
|
|
for.cond: ; preds = %for.body, %entry
|
|
%k.0 = phi i64 [ 0, %entry ], [ %inc, %for.body ]
|
|
%cmp = icmp ult i64 %k.0, %numElem
|
|
br i1 %cmp, label %for.body, label %for.cond.cleanup
|
|
|
|
for.body: ; preds = %for.cond
|
|
%0 = load ptr, ptr %_M_start.i, align 8, !tbaa !3
|
|
%add.ptr.i = getelementptr inbounds double, ptr %0, i64 %k.0
|
|
store double 2.000000e+00, ptr %add.ptr.i, align 8, !tbaa !8
|
|
%inc = add nuw i64 %k.0, 1
|
|
br label %for.cond
|
|
|
|
for.cond.cleanup: ; preds = %for.cond
|
|
ret void
|
|
}
|
|
|
|
!llvm.module.flags = !{!0, !1}
|
|
!llvm.ident = !{!2}
|
|
|
|
!0 = !{i32 1, !"wchar_size", i32 4}
|
|
!1 = !{i32 7, !"uwtable", i32 1}
|
|
!2 = !{!"clang version 15.0.0 (https://github.com/llvm/llvm-project.git fc510998f7c287df2bc1304673e0cd8452d50b31)"}
|
|
!3 = !{!4, !5, i64 0}
|
|
!4 = !{!"_ZTSNSt12_Vector_baseIdSaIdEE17_Vector_impl_dataE", !5, i64 0, !5, i64 8, !5, i64 16}
|
|
!5 = !{!"any pointer", !6, i64 0}
|
|
!6 = !{!"omnipotent char", !7, i64 0}
|
|
!7 = !{!"Simple C++ TBAA"}
|
|
!8 = !{!9, !9, i64 0}
|
|
!9 = !{!"double", !6, i64 0}
|
|
;.
|
|
; O1: [[ANYPTR_TBAA3]] = !{[[META4:![0-9]+]], [[META5:![0-9]+]], i64 0}
|
|
; O1: [[META4]] = !{!"_ZTSNSt12_Vector_baseIdSaIdEE17_Vector_impl_dataE", [[META5]], i64 0, [[META5]], i64 8, [[META5]], i64 16}
|
|
; O1: [[META5]] = !{!"any pointer", [[META6:![0-9]+]], i64 0}
|
|
; O1: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0}
|
|
; O1: [[META7]] = !{!"Simple C++ TBAA"}
|
|
; O1: [[DOUBLE_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0}
|
|
; O1: [[META9]] = !{!"double", [[META6]], i64 0}
|
|
;.
|
|
; O23: [[ANYPTR_TBAA3]] = !{[[META4:![0-9]+]], [[META5:![0-9]+]], i64 0}
|
|
; O23: [[META4]] = !{!"_ZTSNSt12_Vector_baseIdSaIdEE17_Vector_impl_dataE", [[META5]], i64 0, [[META5]], i64 8, [[META5]], i64 16}
|
|
; O23: [[META5]] = !{!"any pointer", [[META6:![0-9]+]], i64 0}
|
|
; O23: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0}
|
|
; O23: [[META7]] = !{!"Simple C++ TBAA"}
|
|
; O23: [[DOUBLE_TBAA8]] = !{[[META9:![0-9]+]], [[META9]], i64 0}
|
|
; O23: [[META9]] = !{!"double", [[META6]], i64 0}
|
|
; O23: [[LOOP10]] = distinct !{[[LOOP10]], [[META11:![0-9]+]], [[META12:![0-9]+]]}
|
|
; O23: [[META11]] = !{!"llvm.loop.isvectorized", i32 1}
|
|
; O23: [[META12]] = !{!"llvm.loop.unroll.runtime.disable"}
|
|
; O23: [[LOOP13]] = distinct !{[[LOOP13]], [[META12]], [[META11]]}
|
|
;.
|