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
132 lines
6.4 KiB
LLVM
132 lines
6.4 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
|
|
; RUN: opt -S -mtriple=amdgcn-- -passes=separate-const-offset-from-gep,slsr,gvn < %s | FileCheck %s
|
|
|
|
target datalayout = "e-p:32:32-p1:64:64-p2:64:64-p3:32:32-p4:64:64-p5:32:32-p24:64:64-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64"
|
|
|
|
|
|
define amdgpu_kernel void @slsr_after_reassociate_global_geps_mubuf_max_offset(ptr addrspace(1) %out, ptr addrspace(1) noalias %arr, i32 %i) {
|
|
; CHECK-LABEL: define amdgpu_kernel void @slsr_after_reassociate_global_geps_mubuf_max_offset(
|
|
; CHECK-SAME: ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) noalias [[ARR:%.*]], i32 [[I:%.*]]) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[I]] to i64
|
|
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr float, ptr addrspace(1) [[ARR]], i64 [[TMP0]]
|
|
; CHECK-NEXT: [[P12:%.*]] = getelementptr i8, ptr addrspace(1) [[TMP1]], i64 4092
|
|
; CHECK-NEXT: [[V11:%.*]] = load i32, ptr addrspace(1) [[P12]], align 4
|
|
; CHECK-NEXT: store i32 [[V11]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[TMP0]], 2
|
|
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr addrspace(1) [[TMP1]], i64 [[TMP2]]
|
|
; CHECK-NEXT: [[P24:%.*]] = getelementptr i8, ptr addrspace(1) [[TMP3]], i64 4092
|
|
; CHECK-NEXT: [[V22:%.*]] = load i32, ptr addrspace(1) [[P24]], align 4
|
|
; CHECK-NEXT: store i32 [[V22]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
bb:
|
|
%i2 = shl nsw i32 %i, 1
|
|
%j1 = add nsw i32 %i, 1023
|
|
%tmp = sext i32 %j1 to i64
|
|
%p1 = getelementptr inbounds float, ptr addrspace(1) %arr, i64 %tmp
|
|
%v11 = load i32, ptr addrspace(1) %p1, align 4
|
|
store i32 %v11, ptr addrspace(1) %out, align 4
|
|
|
|
%j2 = add nsw i32 %i2, 1023
|
|
%tmp5 = sext i32 %j2 to i64
|
|
%p2 = getelementptr inbounds float, ptr addrspace(1) %arr, i64 %tmp5
|
|
%v22 = load i32, ptr addrspace(1) %p2, align 4
|
|
store i32 %v22, ptr addrspace(1) %out, align 4
|
|
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @slsr_after_reassociate_global_geps_over_mubuf_max_offset(ptr addrspace(1) %out, ptr addrspace(1) noalias %arr, i32 %i) {
|
|
; CHECK-LABEL: define amdgpu_kernel void @slsr_after_reassociate_global_geps_over_mubuf_max_offset(
|
|
; CHECK-SAME: ptr addrspace(1) [[OUT:%.*]], ptr addrspace(1) noalias [[ARR:%.*]], i32 [[I:%.*]]) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: [[J1:%.*]] = add nsw i32 [[I]], 1024
|
|
; CHECK-NEXT: [[TMP:%.*]] = sext i32 [[J1]] to i64
|
|
; CHECK-NEXT: [[P1:%.*]] = getelementptr inbounds float, ptr addrspace(1) [[ARR]], i64 [[TMP]]
|
|
; CHECK-NEXT: [[V11:%.*]] = load i32, ptr addrspace(1) [[P1]], align 4
|
|
; CHECK-NEXT: store i32 [[V11]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: [[OFFSET:%.*]] = sext i32 [[I]] to i64
|
|
; CHECK-NEXT: [[TMP5:%.*]] = shl i64 [[OFFSET]], 2
|
|
; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[P1]], i64 [[TMP5]]
|
|
; CHECK-NEXT: [[V22:%.*]] = load i32, ptr addrspace(1) [[P2]], align 4
|
|
; CHECK-NEXT: store i32 [[V22]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
bb:
|
|
%i2 = shl nsw i32 %i, 1
|
|
%j1 = add nsw i32 %i, 1024
|
|
%tmp = sext i32 %j1 to i64
|
|
%p1 = getelementptr inbounds float, ptr addrspace(1) %arr, i64 %tmp
|
|
%v11 = load i32, ptr addrspace(1) %p1, align 4
|
|
store i32 %v11, ptr addrspace(1) %out, align 4
|
|
|
|
%j2 = add nsw i32 %i2, 1024
|
|
%tmp5 = sext i32 %j2 to i64
|
|
%p2 = getelementptr inbounds float, ptr addrspace(1) %arr, i64 %tmp5
|
|
%v22 = load i32, ptr addrspace(1) %p2, align 4
|
|
store i32 %v22, ptr addrspace(1) %out, align 4
|
|
|
|
ret void
|
|
}
|
|
|
|
|
|
define amdgpu_kernel void @slsr_after_reassociate_lds_geps_ds_max_offset(ptr addrspace(1) %out, ptr addrspace(3) noalias %arr, i32 %i) {
|
|
; CHECK-LABEL: define amdgpu_kernel void @slsr_after_reassociate_lds_geps_ds_max_offset(
|
|
; CHECK-SAME: ptr addrspace(1) [[OUT:%.*]], ptr addrspace(3) noalias [[ARR:%.*]], i32 [[I:%.*]]) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr float, ptr addrspace(3) [[ARR]], i32 [[I]]
|
|
; CHECK-NEXT: [[P12:%.*]] = getelementptr i8, ptr addrspace(3) [[TMP0]], i32 65532
|
|
; CHECK-NEXT: [[V11:%.*]] = load i32, ptr addrspace(3) [[P12]], align 4
|
|
; CHECK-NEXT: store i32 [[V11]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[I]], 2
|
|
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr addrspace(3) [[TMP0]], i32 [[TMP1]]
|
|
; CHECK-NEXT: [[P24:%.*]] = getelementptr i8, ptr addrspace(3) [[TMP2]], i32 65532
|
|
; CHECK-NEXT: [[V22:%.*]] = load i32, ptr addrspace(3) [[P24]], align 4
|
|
; CHECK-NEXT: store i32 [[V22]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
bb:
|
|
%i2 = shl nsw i32 %i, 1
|
|
%j1 = add nsw i32 %i, 16383
|
|
%p1 = getelementptr inbounds float, ptr addrspace(3) %arr, i32 %j1
|
|
%v11 = load i32, ptr addrspace(3) %p1, align 4
|
|
store i32 %v11, ptr addrspace(1) %out, align 4
|
|
|
|
%j2 = add nsw i32 %i2, 16383
|
|
%p2 = getelementptr inbounds float, ptr addrspace(3) %arr, i32 %j2
|
|
%v22 = load i32, ptr addrspace(3) %p2, align 4
|
|
store i32 %v22, ptr addrspace(1) %out, align 4
|
|
|
|
ret void
|
|
}
|
|
|
|
define amdgpu_kernel void @slsr_after_reassociate_lds_geps_over_ds_max_offset(ptr addrspace(1) %out, ptr addrspace(3) noalias %arr, i32 %i) {
|
|
; CHECK-LABEL: define amdgpu_kernel void @slsr_after_reassociate_lds_geps_over_ds_max_offset(
|
|
; CHECK-SAME: ptr addrspace(1) [[OUT:%.*]], ptr addrspace(3) noalias [[ARR:%.*]], i32 [[I:%.*]]) {
|
|
; CHECK-NEXT: bb:
|
|
; CHECK-NEXT: [[J1:%.*]] = add nsw i32 [[I]], 16384
|
|
; CHECK-NEXT: [[P1:%.*]] = getelementptr inbounds float, ptr addrspace(3) [[ARR]], i32 [[J1]]
|
|
; CHECK-NEXT: [[V11:%.*]] = load i32, ptr addrspace(3) [[P1]], align 4
|
|
; CHECK-NEXT: store i32 [[V11]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: [[J2:%.*]] = shl i32 [[I]], 2
|
|
; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i8, ptr addrspace(3) [[P1]], i32 [[J2]]
|
|
; CHECK-NEXT: [[V22:%.*]] = load i32, ptr addrspace(3) [[P2]], align 4
|
|
; CHECK-NEXT: store i32 [[V22]], ptr addrspace(1) [[OUT]], align 4
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
bb:
|
|
%i2 = shl nsw i32 %i, 1
|
|
%j1 = add nsw i32 %i, 16384
|
|
%p1 = getelementptr inbounds float, ptr addrspace(3) %arr, i32 %j1
|
|
%v11 = load i32, ptr addrspace(3) %p1, align 4
|
|
store i32 %v11, ptr addrspace(1) %out, align 4
|
|
|
|
%j2 = add nsw i32 %i2, 16384
|
|
%p2 = getelementptr inbounds float, ptr addrspace(3) %arr, i32 %j2
|
|
%v22 = load i32, ptr addrspace(3) %p2, align 4
|
|
store i32 %v22, ptr addrspace(1) %out, align 4
|
|
|
|
ret void
|
|
}
|