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
162 lines
6.6 KiB
LLVM
162 lines
6.6 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals
|
|
; RUN: opt < %s -passes=instcombine -S | FileCheck %s
|
|
|
|
; load(addrspacecast(select(A, B))) --> select(load(addrspacecast(A)), load(addrspacecast(B)))
|
|
|
|
define i32 @load_addrspacecast_select_alloca(i1 %cond) !prof !0 {
|
|
; CHECK-LABEL: @load_addrspacecast_select_alloca(
|
|
; CHECK-NEXT: [[RESULT:%.*]] = select i1 [[COND:%.*]], i32 42, i32 100, !prof [[PROF1:![0-9]+]]
|
|
; CHECK-NEXT: ret i32 [[RESULT]]
|
|
;
|
|
%a = alloca i32, align 4
|
|
%b = alloca i32, align 4
|
|
store i32 42, ptr %a, align 4
|
|
store i32 100, ptr %b, align 4
|
|
%sel = select i1 %cond, ptr %a, ptr %b, !prof !1
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load i32, ptr addrspace(5) %cast, align 4
|
|
ret i32 %load
|
|
}
|
|
|
|
define <2 x i32> @load_addrspacecast_select_alloca_vec(i1 %cond) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_alloca_vec(
|
|
; CHECK-NEXT: [[RESULT:%.*]] = select i1 [[COND:%.*]], <2 x i32> <i32 42, i32 43>, <2 x i32> <i32 100, i32 101>
|
|
; CHECK-NEXT: ret <2 x i32> [[RESULT]]
|
|
;
|
|
%a = alloca <2 x i32>, align 8
|
|
%b = alloca <2 x i32>, align 8
|
|
store <2 x i32> <i32 42, i32 43>, ptr %a, align 8
|
|
store <2 x i32> <i32 100, i32 101>, ptr %b, align 8
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load <2 x i32>, ptr addrspace(5) %cast, align 8
|
|
ret <2 x i32> %load
|
|
}
|
|
|
|
; Negative tests:
|
|
|
|
|
|
define i32 @load_addrspacecast_select_multi_use(i1 %cond, ptr %out) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_multi_use(
|
|
; CHECK-NEXT: [[A:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: [[B:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: store i32 42, ptr [[A]], align 4
|
|
; CHECK-NEXT: store i32 100, ptr [[B]], align 4
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A]], ptr [[B]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr addrspace(5) [[CAST]], align 4
|
|
; CHECK-NEXT: store ptr [[SEL]], ptr [[OUT:%.*]], align 8
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%a = alloca i32, align 4
|
|
%b = alloca i32, align 4
|
|
store i32 42, ptr %a, align 4
|
|
store i32 100, ptr %b, align 4
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load i32, ptr addrspace(5) %cast, align 4
|
|
; Extra use of the select prevents transformation
|
|
store ptr %sel, ptr %out, align 8
|
|
ret i32 %load
|
|
}
|
|
|
|
define i32 @load_addrspacecast_multi_use(i1 %cond, ptr addrspace(5) %out) {
|
|
; CHECK-LABEL: @load_addrspacecast_multi_use(
|
|
; CHECK-NEXT: [[A:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: [[B:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: store i32 42, ptr [[A]], align 4
|
|
; CHECK-NEXT: store i32 100, ptr [[B]], align 4
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A]], ptr [[B]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr addrspace(5) [[CAST]], align 4
|
|
; CHECK-NEXT: store ptr addrspace(5) [[CAST]], ptr addrspace(5) [[OUT:%.*]], align 8
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%a = alloca i32, align 4
|
|
%b = alloca i32, align 4
|
|
store i32 42, ptr %a, align 4
|
|
store i32 100, ptr %b, align 4
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load i32, ptr addrspace(5) %cast, align 4
|
|
; Extra use of the cast prevents transformation
|
|
store ptr addrspace(5) %cast, ptr addrspace(5) %out, align 8
|
|
ret i32 %load
|
|
}
|
|
|
|
define i32 @load_addrspacecast_select_on_args(i1 %cond, ptr %a, ptr %b) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_on_args(
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A:%.*]], ptr [[B:%.*]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr addrspace(5) [[CAST]], align 4
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load i32, ptr addrspace(5) %cast, align 4
|
|
ret i32 %load
|
|
}
|
|
|
|
; dereferenceable is not enough to trigger the transformation
|
|
define i32 @load_addrspacecast_select_on_args_dereferenceable(i1 %cond, ptr dereferenceable(4) %a, ptr dereferenceable(4) %b) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_on_args_dereferenceable(
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A:%.*]], ptr [[B:%.*]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr addrspace(5) [[CAST]], align 4
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load i32, ptr addrspace(5) %cast, align 4
|
|
ret i32 %load
|
|
}
|
|
|
|
define i32 @load_addrspacecast_select_volatile(i1 %cond) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_volatile(
|
|
; CHECK-NEXT: [[A:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: [[B:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: store i32 42, ptr [[A]], align 4
|
|
; CHECK-NEXT: store i32 100, ptr [[B]], align 4
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A]], ptr [[B]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load volatile i32, ptr addrspace(5) [[CAST]], align 4
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%a = alloca i32, align 4
|
|
%b = alloca i32, align 4
|
|
store i32 42, ptr %a, align 4
|
|
store i32 100, ptr %b, align 4
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load volatile i32, ptr addrspace(5) %cast, align 4
|
|
ret i32 %load
|
|
}
|
|
|
|
define i32 @load_addrspacecast_select_atomic(i1 %cond) {
|
|
; CHECK-LABEL: @load_addrspacecast_select_atomic(
|
|
; CHECK-NEXT: [[A:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: [[B:%.*]] = alloca i32, align 4
|
|
; CHECK-NEXT: store i32 42, ptr [[A]], align 4
|
|
; CHECK-NEXT: store i32 100, ptr [[B]], align 4
|
|
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[COND:%.*]], ptr [[A]], ptr [[B]]
|
|
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr [[SEL]] to ptr addrspace(5)
|
|
; CHECK-NEXT: [[LOAD:%.*]] = load atomic i32, ptr addrspace(5) [[CAST]] seq_cst, align 4
|
|
; CHECK-NEXT: ret i32 [[LOAD]]
|
|
;
|
|
%a = alloca i32, align 4
|
|
%b = alloca i32, align 4
|
|
store i32 42, ptr %a, align 4
|
|
store i32 100, ptr %b, align 4
|
|
%sel = select i1 %cond, ptr %a, ptr %b
|
|
%cast = addrspacecast ptr %sel to ptr addrspace(5)
|
|
%load = load atomic i32, ptr addrspace(5) %cast seq_cst, align 4
|
|
ret i32 %load
|
|
}
|
|
|
|
!0 = !{!"function_entry_count", i64 1000}
|
|
!1 = !{!"branch_weights", i32 2, i32 3}
|
|
;.
|
|
; CHECK: [[META0:![0-9]+]] = !{!"function_entry_count", i64 1000}
|
|
; CHECK: [[PROF1]] = !{!"branch_weights", i32 2, i32 3}
|
|
;.
|