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
65 lines
2.3 KiB
LLVM
65 lines
2.3 KiB
LLVM
; RUN: llc -mtriple=hexagon -enable-pipeliner -pipeliner-max-stages=2 < %s -pipeliner-experimental-cg=true | FileCheck %s
|
|
|
|
; Test that we generate the correct offsets after we removed unneeded
|
|
; chain dependences between Phis and generated a better pipeline.
|
|
|
|
; CHECK: loop0(.LBB0_[[LOOP:.]],
|
|
; CHECK: .LBB0_[[LOOP]]:
|
|
; CHECK: = memd([[REG0:(r[0-9]+)]]+#8)
|
|
; CHECK: memd([[REG0]]++#8) =
|
|
; CHECK: }{{[ \t]*}}:endloop0
|
|
|
|
@g0 = common global [400 x i8] zeroinitializer, align 8
|
|
@g1 = common global [400 x i8] zeroinitializer, align 8
|
|
|
|
; Function Attrs: nounwind
|
|
define void @f0() #0 {
|
|
b0:
|
|
br label %b2
|
|
|
|
b1: ; preds = %b2
|
|
ret void
|
|
|
|
b2: ; preds = %b2, %b0
|
|
%v0 = phi ptr [ @g0, %b0 ], [ %v23, %b2 ]
|
|
%v1 = phi ptr [ @g1, %b0 ], [ %v24, %b2 ]
|
|
%v2 = phi i32 [ 0, %b0 ], [ %v21, %b2 ]
|
|
%v4 = load <8 x i8>, ptr %v0, align 8
|
|
%v6 = load <8 x i8>, ptr %v1, align 8
|
|
%v7 = bitcast <8 x i8> %v4 to <2 x i32>
|
|
%v8 = extractelement <2 x i32> %v7, i32 0
|
|
%v9 = extractelement <2 x i32> %v7, i32 1
|
|
%v10 = bitcast <8 x i8> %v6 to <2 x i32>
|
|
%v11 = extractelement <2 x i32> %v10, i32 0
|
|
%v12 = extractelement <2 x i32> %v10, i32 1
|
|
%v13 = tail call i64 @llvm.hexagon.S2.vzxtbh(i32 %v11)
|
|
%v14 = tail call i64 @llvm.hexagon.S2.vzxtbh(i32 %v12)
|
|
%v15 = tail call i64 @llvm.hexagon.M5.vmacbsu(i64 %v13, i32 %v8, i32 117901063)
|
|
%v16 = tail call i64 @llvm.hexagon.M5.vmacbsu(i64 %v14, i32 %v9, i32 117901063)
|
|
%v17 = tail call i32 @llvm.hexagon.S2.vtrunehb(i64 %v15)
|
|
%v18 = tail call i32 @llvm.hexagon.S2.vtrunehb(i64 %v16)
|
|
%v19 = tail call i64 @llvm.hexagon.A2.combinew(i32 %v18, i32 %v17)
|
|
%v20 = bitcast i64 %v19 to <8 x i8>
|
|
store <8 x i8> %v20, ptr %v1, align 8
|
|
%v21 = add nsw i32 %v2, 8
|
|
%v22 = icmp slt i32 %v2, 392
|
|
%v23 = getelementptr i8, ptr %v0, i32 8
|
|
%v24 = getelementptr i8, ptr %v1, i32 8
|
|
br i1 %v22, label %b2, label %b1
|
|
}
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @llvm.hexagon.S2.vzxtbh(i32) #1
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @llvm.hexagon.M5.vmacbsu(i64, i32, i32) #1
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i32 @llvm.hexagon.S2.vtrunehb(i64) #1
|
|
|
|
; Function Attrs: nounwind readnone
|
|
declare i64 @llvm.hexagon.A2.combinew(i32, i32) #1
|
|
|
|
attributes #0 = { nounwind }
|
|
attributes #1 = { nounwind readnone }
|