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
104 lines
6.0 KiB
LLVM
104 lines
6.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph" --version 6
|
|
; RUN: opt %s -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -enable-interleaved-mem-accesses -S | FileCheck %s
|
|
; RUN: opt %s -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -enable-interleaved-mem-accesses -force-target-supports-scalable-vectors=true -scalable-vectorization=preferred -S | FileCheck %s --check-prefix=SCALABLE
|
|
|
|
define void @scalablevf(ptr %dst.start, i8 %a, i8 %b) {
|
|
; CHECK-LABEL: define void @scalablevf(
|
|
; CHECK-SAME: ptr [[DST_START:%.*]], i8 [[A:%.*]], i8 [[B:%.*]]) {
|
|
; CHECK-NEXT: [[ENTRY:.*:]]
|
|
; CHECK-NEXT: br label %[[VECTOR_PH:.*]]
|
|
; CHECK: [[VECTOR_PH]]:
|
|
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i8> poison, i8 [[B]], i64 0
|
|
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i8> [[BROADCAST_SPLATINSERT]], <4 x i8> poison, <4 x i32> zeroinitializer
|
|
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i8> poison, i8 [[A]], i64 0
|
|
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i8> [[BROADCAST_SPLATINSERT1]], <4 x i8> poison, <4 x i32> zeroinitializer
|
|
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; CHECK: [[VECTOR_BODY]]:
|
|
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
|
|
; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST_START]], i64 [[OFFSET_IDX]]
|
|
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[NEXT_GEP]], align 1
|
|
; CHECK-NEXT: [[TMP0:%.*]] = mul nuw <4 x i8> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
|
|
; CHECK-NEXT: [[TMP1:%.*]] = udiv <4 x i8> [[TMP0]], splat (i8 -1)
|
|
; CHECK-NEXT: [[TMP2:%.*]] = add <4 x i8> [[BROADCAST_SPLAT2]], [[TMP1]]
|
|
; CHECK-NEXT: store <4 x i8> [[TMP2]], ptr [[NEXT_GEP]], align 1
|
|
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 1
|
|
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 64
|
|
; CHECK-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; CHECK: [[MIDDLE_BLOCK]]:
|
|
; CHECK-NEXT: br label %[[EXIT:.*]]
|
|
; CHECK: [[EXIT]]:
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
; SCALABLE-LABEL: define void @scalablevf(
|
|
; SCALABLE-SAME: ptr [[DST_START:%.*]], i8 [[A:%.*]], i8 [[B:%.*]]) {
|
|
; SCALABLE-NEXT: [[ENTRY:.*:]]
|
|
; SCALABLE-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
|
|
; SCALABLE-NEXT: [[TMP1:%.*]] = shl nuw i64 [[TMP0]], 2
|
|
; SCALABLE-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 64, [[TMP1]]
|
|
; SCALABLE-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
|
|
; SCALABLE: [[VECTOR_PH]]:
|
|
; SCALABLE-NEXT: [[N_MOD_VF:%.*]] = urem i64 64, [[TMP0]]
|
|
; SCALABLE-NEXT: [[N_VEC:%.*]] = sub i64 64, [[N_MOD_VF]]
|
|
; SCALABLE-NEXT: [[TMP5:%.*]] = shl i64 [[N_VEC]], 2
|
|
; SCALABLE-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[DST_START]], i64 [[TMP5]]
|
|
; SCALABLE-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i8> poison, i8 [[B]], i64 0
|
|
; SCALABLE-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i8> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i8> poison, <vscale x 4 x i32> zeroinitializer
|
|
; SCALABLE-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i8> poison, i8 [[A]], i64 0
|
|
; SCALABLE-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i8> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i8> poison, <vscale x 4 x i32> zeroinitializer
|
|
; SCALABLE-NEXT: br label %[[VECTOR_BODY:.*]]
|
|
; SCALABLE: [[VECTOR_BODY]]:
|
|
; SCALABLE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
|
|
; SCALABLE-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
|
|
; SCALABLE-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST_START]], i64 [[OFFSET_IDX]]
|
|
; SCALABLE-NEXT: [[TMP10:%.*]] = load <vscale x 4 x i8>, ptr [[NEXT_GEP]], align 1
|
|
; SCALABLE-NEXT: [[TMP20:%.*]] = mul nuw <vscale x 4 x i8> [[TMP10]], [[BROADCAST_SPLAT]]
|
|
; SCALABLE-NEXT: [[TMP21:%.*]] = udiv <vscale x 4 x i8> [[TMP20]], splat (i8 -1)
|
|
; SCALABLE-NEXT: [[TMP22:%.*]] = add <vscale x 4 x i8> [[BROADCAST_SPLAT2]], [[TMP21]]
|
|
; SCALABLE-NEXT: store <vscale x 4 x i8> [[TMP22]], ptr [[NEXT_GEP]], align 1
|
|
; SCALABLE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP0]]
|
|
; SCALABLE-NEXT: [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
|
|
; SCALABLE-NEXT: br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
|
|
; SCALABLE: [[MIDDLE_BLOCK]]:
|
|
; SCALABLE-NEXT: [[CMP_N:%.*]] = icmp eq i64 64, [[N_VEC]]
|
|
; SCALABLE-NEXT: br i1 [[CMP_N]], [[EXIT:label %.*]], label %[[SCALAR_PH]]
|
|
; SCALABLE: [[SCALAR_PH]]:
|
|
;
|
|
entry:
|
|
br label %loop
|
|
|
|
loop:
|
|
%iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
|
|
%dst = phi ptr [ %dst.start, %entry ], [ %dst.next, %loop ]
|
|
%dst.next = getelementptr inbounds nuw i8, ptr %dst, i64 4
|
|
%load.dst = load i8, ptr %dst, align 1
|
|
%mul.dst.0 = mul nuw i8 %load.dst, %b
|
|
%udiv.0 = udiv i8 %mul.dst.0, 255
|
|
%val.0 = add i8 %a, %udiv.0
|
|
store i8 %val.0, ptr %dst, align 1
|
|
%gep.dst.1 = getelementptr inbounds nuw i8, ptr %dst, i64 1
|
|
%load.dst.1 = load i8, ptr %gep.dst.1, align 1
|
|
%mul.dst.1 = mul nuw i8 %load.dst.1, %b
|
|
%udiv.1 = udiv i8 %mul.dst.1, 255
|
|
%val.1 = add i8 %a, %udiv.1
|
|
store i8 %val.1, ptr %gep.dst.1, align 1
|
|
%gep.dst.2 = getelementptr inbounds nuw i8, ptr %dst, i64 2
|
|
%load.dst.2 = load i8, ptr %gep.dst.2, align 1
|
|
%mul.dst.2 = mul nuw i8 %load.dst.2, %b
|
|
%udiv.2 = udiv i8 %mul.dst.2, 255
|
|
%val.2 = add i8 %a, %udiv.2
|
|
store i8 %val.2, ptr %gep.dst.2, align 1
|
|
%gep.dst.3 = getelementptr inbounds nuw i8, ptr %dst, i64 3
|
|
%load.dst.3 = load i8, ptr %gep.dst.3, align 1
|
|
%mul.dst.3 = mul nuw i8 %load.dst.3, %b
|
|
%udiv.3 = udiv i8 %mul.dst.3, 255
|
|
%val.3 = add i8 %a, %udiv.3
|
|
store i8 %val.3, ptr %gep.dst.3, align 1
|
|
%iv.next = add i64 %iv, 4
|
|
%exit.cond = icmp eq i64 %iv.next, 256
|
|
br i1 %exit.cond, label %exit, label %loop
|
|
|
|
exit:
|
|
ret void
|
|
}
|