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
75 lines
3.6 KiB
LLVM
75 lines
3.6 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt %s -passes='function(scalarizer<load-store>)' -S | FileCheck %s
|
|
|
|
; This verifies that the order of extract element instructions is
|
|
; deterministic. In the past we could end up with different results depending
|
|
; on the compiler used (due to argument evaluation order being undefined in
|
|
; C++). The order of the extracts is not really important for correctness of
|
|
; the result, but when debugging and creating test cases it is helpful if we
|
|
; get the same out put regardless of which compiler we use when building the
|
|
; compiler.
|
|
|
|
define <2 x i32> @test1(i1 %b, <2 x i32> %i, <2 x i32> %j) {
|
|
; CHECK-LABEL: @test1(
|
|
; CHECK-NEXT: [[I_I0:%.*]] = extractelement <2 x i32> [[I:%.*]], i64 0
|
|
; CHECK-NEXT: [[J_I0:%.*]] = extractelement <2 x i32> [[J:%.*]], i64 0
|
|
; CHECK-NEXT: [[RES_I0:%.*]] = select i1 [[B:%.*]], i32 [[I_I0]], i32 [[J_I0]]
|
|
; CHECK-NEXT: [[I_I1:%.*]] = extractelement <2 x i32> [[I]], i64 1
|
|
; CHECK-NEXT: [[J_I1:%.*]] = extractelement <2 x i32> [[J]], i64 1
|
|
; CHECK-NEXT: [[RES_I1:%.*]] = select i1 [[B]], i32 [[I_I1]], i32 [[J_I1]]
|
|
; CHECK-NEXT: [[RES_UPTO0:%.*]] = insertelement <2 x i32> poison, i32 [[RES_I0]], i64 0
|
|
; CHECK-NEXT: [[RES:%.*]] = insertelement <2 x i32> [[RES_UPTO0]], i32 [[RES_I1]], i64 1
|
|
; CHECK-NEXT: ret <2 x i32> [[RES]]
|
|
;
|
|
%res = select i1 %b, <2 x i32> %i, <2 x i32> %j
|
|
ret <2 x i32> %res
|
|
}
|
|
|
|
define <2 x i32> @test2(<2 x i1> %b, <2 x i32> %i, <2 x i32> %j) {
|
|
; CHECK-LABEL: @test2(
|
|
; CHECK-NEXT: [[B_I0:%.*]] = extractelement <2 x i1> [[B:%.*]], i64 0
|
|
; CHECK-NEXT: [[I_I0:%.*]] = extractelement <2 x i32> [[I:%.*]], i64 0
|
|
; CHECK-NEXT: [[J_I0:%.*]] = extractelement <2 x i32> [[J:%.*]], i64 0
|
|
; CHECK-NEXT: [[RES_I0:%.*]] = select i1 [[B_I0]], i32 [[I_I0]], i32 [[J_I0]]
|
|
; CHECK-NEXT: [[B_I1:%.*]] = extractelement <2 x i1> [[B]], i64 1
|
|
; CHECK-NEXT: [[I_I1:%.*]] = extractelement <2 x i32> [[I]], i64 1
|
|
; CHECK-NEXT: [[J_I1:%.*]] = extractelement <2 x i32> [[J]], i64 1
|
|
; CHECK-NEXT: [[RES_I1:%.*]] = select i1 [[B_I1]], i32 [[I_I1]], i32 [[J_I1]]
|
|
; CHECK-NEXT: [[RES_UPTO0:%.*]] = insertelement <2 x i32> poison, i32 [[RES_I0]], i64 0
|
|
; CHECK-NEXT: [[RES:%.*]] = insertelement <2 x i32> [[RES_UPTO0]], i32 [[RES_I1]], i64 1
|
|
; CHECK-NEXT: ret <2 x i32> [[RES]]
|
|
;
|
|
%res = select <2 x i1> %b, <2 x i32> %i, <2 x i32> %j
|
|
ret <2 x i32> %res
|
|
}
|
|
|
|
define <2 x i32> @test3(<2 x i32> %i, <2 x i32> %j) {
|
|
; CHECK-LABEL: @test3(
|
|
; CHECK-NEXT: [[I_I0:%.*]] = extractelement <2 x i32> [[I:%.*]], i64 0
|
|
; CHECK-NEXT: [[J_I0:%.*]] = extractelement <2 x i32> [[J:%.*]], i64 0
|
|
; CHECK-NEXT: [[RES_I0:%.*]] = add nuw nsw i32 [[I_I0]], [[J_I0]]
|
|
; CHECK-NEXT: [[I_I1:%.*]] = extractelement <2 x i32> [[I]], i64 1
|
|
; CHECK-NEXT: [[J_I1:%.*]] = extractelement <2 x i32> [[J]], i64 1
|
|
; CHECK-NEXT: [[RES_I1:%.*]] = add nuw nsw i32 [[I_I1]], [[J_I1]]
|
|
; CHECK-NEXT: [[RES_UPTO0:%.*]] = insertelement <2 x i32> poison, i32 [[RES_I0]], i64 0
|
|
; CHECK-NEXT: [[RES:%.*]] = insertelement <2 x i32> [[RES_UPTO0]], i32 [[RES_I1]], i64 1
|
|
; CHECK-NEXT: ret <2 x i32> [[RES]]
|
|
;
|
|
%res = add nuw nsw <2 x i32> %i, %j
|
|
ret <2 x i32> %res
|
|
}
|
|
|
|
define void @test4(ptr %ptr, <2 x i32> %val) {
|
|
; CHECK-LABEL: @test4(
|
|
; CHECK-NEXT: [[VAL_I0:%.*]] = extractelement <2 x i32> [[VAL:%.*]], i64 0
|
|
; CHECK-NEXT: store i32 [[VAL_I0]], ptr [[PTR:%.*]], align 8
|
|
; CHECK-NEXT: [[VAL_I1:%.*]] = extractelement <2 x i32> [[VAL]], i64 1
|
|
; CHECK-NEXT: [[PTR_I1:%.*]] = getelementptr i32, ptr [[PTR]], i32 1
|
|
; CHECK-NEXT: store i32 [[VAL_I1]], ptr [[PTR_I1]], align 4
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
store <2 x i32> %val, ptr %ptr
|
|
ret void
|
|
}
|
|
|