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
155 lines
7.8 KiB
LLVM
155 lines
7.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
|
|
; RUN: opt -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=2 -S %s | FileCheck %s
|
|
|
|
define void @preserve_inbounds_through_noop_gep(ptr %p) {
|
|
; CHECK-LABEL: define void @preserve_inbounds_through_noop_gep(
|
|
; CHECK-SAME: ptr [[P:%.*]]) {
|
|
; CHECK-NEXT: [[VECTOR_BODY:.*:]]
|
|
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
|
|
; CHECK: [[VECTOR_PH]]:
|
|
; CHECK-NEXT: br label %[[VECTOR_BODY1:.*]]
|
|
; CHECK: [[VECTOR_BODY1]]:
|
|
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY1]] ]
|
|
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[P]], i32 [[INDEX]]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i64 0
|
|
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
|
|
; CHECK-NEXT: store <4 x i32> zeroinitializer, ptr [[TMP1]], align 4
|
|
; CHECK-NEXT: store <4 x i32> zeroinitializer, ptr [[TMP2]], align 4
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
|
|
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
|
|
; CHECK-NEXT: br i1 [[TMP3]], label %[[SCALAR_PH:.*]], label %[[VECTOR_BODY1]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; CHECK: [[SCALAR_PH]]:
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ]
|
|
%gep = getelementptr inbounds i32, ptr %p, i32 %iv
|
|
%noop = getelementptr i8, ptr %gep, i64 0
|
|
store i32 0, ptr %noop
|
|
%iv.next = add nsw i32 %iv, 1
|
|
%ec = icmp eq i32 %iv.next, 1000
|
|
br i1 %ec, label %exit, label %loop
|
|
|
|
exit:
|
|
ret void
|
|
}
|
|
|
|
define void @bitcast_in_ptr_chain(ptr %src, ptr noalias %dst) {
|
|
; CHECK-LABEL: define void @bitcast_in_ptr_chain(
|
|
; CHECK-SAME: ptr [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
|
|
; CHECK: [[VECTOR_PH]]:
|
|
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; CHECK: [[VECTOR_BODY]]:
|
|
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds double, ptr [[SRC]], i64 [[INDEX]]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast ptr [[TMP0]] to ptr
|
|
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[TMP1]], i64 4
|
|
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP1]], align 8
|
|
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x double>, ptr [[TMP2]], align 8
|
|
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
|
|
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[TMP3]], i64 4
|
|
; CHECK-NEXT: store <4 x double> [[WIDE_LOAD]], ptr [[TMP3]], align 8
|
|
; CHECK-NEXT: store <4 x double> [[WIDE_LOAD1]], ptr [[TMP4]], align 8
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
|
|
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
|
|
; CHECK-NEXT: br i1 [[TMP5]], label %[[SCALAR_PH:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
|
|
; CHECK: [[SCALAR_PH]]:
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
|
|
%gep.src = getelementptr inbounds double, ptr %src, i64 %iv
|
|
%bc = bitcast ptr %gep.src to ptr
|
|
%val = load double, ptr %bc, align 8
|
|
%gep.dst = getelementptr inbounds double, ptr %dst, i64 %iv
|
|
store double %val, ptr %gep.dst, align 8
|
|
%iv.next = add i64 %iv, 1
|
|
%ec = icmp eq i64 %iv.next, 1000
|
|
br i1 %ec, label %exit, label %loop
|
|
|
|
exit:
|
|
ret void
|
|
}
|
|
|
|
define void @addrspacecast_in_ptr_chain(ptr addrspace(1) %src, ptr noalias %dst) {
|
|
; CHECK-LABEL: define void @addrspacecast_in_ptr_chain(
|
|
; CHECK-SAME: ptr addrspace(1) [[SRC:%.*]], ptr noalias [[DST:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
|
|
; CHECK: [[VECTOR_PH]]:
|
|
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; CHECK: [[VECTOR_BODY]]:
|
|
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[STEP_ADD:%.*]] = add nuw <4 x i64> [[VEC_IND]], splat (i64 4)
|
|
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds double, ptr addrspace(1) [[SRC]], <4 x i64> [[VEC_IND]]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr addrspace(1) [[SRC]], <4 x i64> [[STEP_ADD]]
|
|
; CHECK-NEXT: [[TMP2:%.*]] = addrspacecast <4 x ptr addrspace(1)> [[TMP0]] to <4 x ptr>
|
|
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x ptr> [[TMP2]], i64 0
|
|
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x ptr> [[TMP2]], i64 1
|
|
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x ptr> [[TMP2]], i64 2
|
|
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x ptr> [[TMP2]], i64 3
|
|
; CHECK-NEXT: [[TMP7:%.*]] = addrspacecast <4 x ptr addrspace(1)> [[TMP1]] to <4 x ptr>
|
|
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <4 x ptr> [[TMP7]], i64 0
|
|
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x ptr> [[TMP7]], i64 1
|
|
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x ptr> [[TMP7]], i64 2
|
|
; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x ptr> [[TMP7]], i64 3
|
|
; CHECK-NEXT: [[TMP12:%.*]] = load double, ptr [[TMP3]], align 8
|
|
; CHECK-NEXT: [[TMP13:%.*]] = load double, ptr [[TMP4]], align 8
|
|
; CHECK-NEXT: [[TMP14:%.*]] = load double, ptr [[TMP5]], align 8
|
|
; CHECK-NEXT: [[TMP15:%.*]] = load double, ptr [[TMP6]], align 8
|
|
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x double> poison, double [[TMP12]], i32 0
|
|
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x double> [[TMP16]], double [[TMP13]], i32 1
|
|
; CHECK-NEXT: [[TMP18:%.*]] = insertelement <4 x double> [[TMP17]], double [[TMP14]], i32 2
|
|
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x double> [[TMP18]], double [[TMP15]], i32 3
|
|
; CHECK-NEXT: [[TMP20:%.*]] = load double, ptr [[TMP8]], align 8
|
|
; CHECK-NEXT: [[TMP21:%.*]] = load double, ptr [[TMP9]], align 8
|
|
; CHECK-NEXT: [[TMP22:%.*]] = load double, ptr [[TMP10]], align 8
|
|
; CHECK-NEXT: [[TMP23:%.*]] = load double, ptr [[TMP11]], align 8
|
|
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x double> poison, double [[TMP20]], i32 0
|
|
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x double> [[TMP24]], double [[TMP21]], i32 1
|
|
; CHECK-NEXT: [[TMP26:%.*]] = insertelement <4 x double> [[TMP25]], double [[TMP22]], i32 2
|
|
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x double> [[TMP26]], double [[TMP23]], i32 3
|
|
; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
|
|
; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds double, ptr [[TMP28]], i64 4
|
|
; CHECK-NEXT: store <4 x double> [[TMP19]], ptr [[TMP28]], align 8
|
|
; CHECK-NEXT: store <4 x double> [[TMP27]], ptr [[TMP29]], align 8
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
|
|
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[STEP_ADD]], splat (i64 4)
|
|
; CHECK-NEXT: [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
|
|
; CHECK-NEXT: br i1 [[TMP30]], label %[[SCALAR_PH:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
|
|
; CHECK: [[SCALAR_PH]]:
|
|
; CHECK-NEXT: br label %[[LOOP:.*]]
|
|
; CHECK: [[LOOP]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
|
|
%gep.src = getelementptr inbounds double, ptr addrspace(1) %src, i64 %iv
|
|
%ac = addrspacecast ptr addrspace(1) %gep.src to ptr
|
|
%val = load double, ptr %ac, align 8
|
|
%gep.dst = getelementptr inbounds double, ptr %dst, i64 %iv
|
|
store double %val, ptr %gep.dst, align 8
|
|
%iv.next = add i64 %iv, 1
|
|
%ec = icmp eq i64 %iv.next, 1000
|
|
br i1 %ec, label %exit, label %loop
|
|
|
|
exit:
|
|
ret void
|
|
}
|