Files
TheRock/compiler/amd-llvm/llvm/test/Transforms/LowerMatrixIntrinsics/unary.ll
T
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
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

251 lines
13 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s
define void @fneg_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @fneg_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[IN:%.*]], align 16
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds float, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x float>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = fneg <2 x float> [[COL_LOAD]]
; CHECK-NEXT: [[TMP2:%.*]] = fneg <2 x float> [[COL_LOAD1]]
; CHECK-NEXT: store <2 x float> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x float> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x float>, ptr %in
%op = fneg <4 x float> %inv
call void @llvm.matrix.column.major.store(<4 x float> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @trunc_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @trunc_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i64, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = trunc <2 x i64> [[COL_LOAD]] to <2 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = trunc <2 x i64> [[COL_LOAD1]] to <2 x i32>
; CHECK-NEXT: store <2 x i32> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i32> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x i64>, ptr %in
%op = trunc <4 x i64> %inv to <4 x i32>
call void @llvm.matrix.column.major.store(<4 x i32> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @zext_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @zext_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i16>, ptr [[IN:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i16, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i16>, ptr [[VEC_GEP]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = zext <2 x i16> [[COL_LOAD]] to <2 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = zext <2 x i16> [[COL_LOAD1]] to <2 x i32>
; CHECK-NEXT: store <2 x i32> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i32> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x i16>, ptr %in
%op = zext <4 x i16> %inv to <4 x i32>
call void @llvm.matrix.column.major.store(<4 x i32> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @sext_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @sext_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i8>, ptr [[IN:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i8>, ptr [[VEC_GEP]], align 2
; CHECK-NEXT: [[TMP1:%.*]] = sext <2 x i8> [[COL_LOAD]] to <2 x i16>
; CHECK-NEXT: [[TMP2:%.*]] = sext <2 x i8> [[COL_LOAD1]] to <2 x i16>
; CHECK-NEXT: store <2 x i16> [[TMP1]], ptr [[OUT:%.*]], align 2
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i16, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i16> [[TMP2]], ptr [[VEC_GEP2]], align 2
; CHECK-NEXT: ret void
;
%inv = load <4 x i8>, ptr %in
%op = sext <4 x i8> %inv to <4 x i16>
call void @llvm.matrix.column.major.store(<4 x i16> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @fptoui_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @fptoui_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[IN:%.*]], align 16
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds float, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x float>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = fptoui <2 x float> [[COL_LOAD]] to <2 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = fptoui <2 x float> [[COL_LOAD1]] to <2 x i32>
; CHECK-NEXT: store <2 x i32> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i32> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x float>, ptr %in
%op = fptoui <4 x float> %inv to <4 x i32>
call void @llvm.matrix.column.major.store(<4 x i32> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @fptosi_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @fptosi_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[IN:%.*]], align 16
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds float, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x float>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = fptosi <2 x float> [[COL_LOAD]] to <2 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = fptosi <2 x float> [[COL_LOAD1]] to <2 x i32>
; CHECK-NEXT: store <2 x i32> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i32> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x float>, ptr %in
%op = fptosi <4 x float> %inv to <4 x i32>
call void @llvm.matrix.column.major.store(<4 x i32> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @uitofp_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @uitofp_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i64, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = uitofp <2 x i64> [[COL_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP2:%.*]] = uitofp <2 x i64> [[COL_LOAD1]] to <2 x double>
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[OUT:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds double, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x double> [[TMP2]], ptr [[VEC_GEP2]], align 8
; CHECK-NEXT: ret void
;
%inv = load <4 x i64>, ptr %in
%op = uitofp <4 x i64> %inv to <4 x double>
call void @llvm.matrix.column.major.store(<4 x double> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @sitofp_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @sitofp_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i64, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = sitofp <2 x i64> [[COL_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP2:%.*]] = sitofp <2 x i64> [[COL_LOAD1]] to <2 x double>
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[OUT:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds double, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x double> [[TMP2]], ptr [[VEC_GEP2]], align 8
; CHECK-NEXT: ret void
;
%inv = load <4 x i64>, ptr %in
%op = sitofp <4 x i64> %inv to <4 x double>
call void @llvm.matrix.column.major.store(<4 x double> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @fptrunc_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @fptrunc_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds double, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = fptrunc nnan <2 x double> [[COL_LOAD]] to <2 x float>
; CHECK-NEXT: [[TMP2:%.*]] = fptrunc nnan <2 x double> [[COL_LOAD1]] to <2 x float>
; CHECK-NEXT: store <2 x float> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x float> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x double>, ptr %in
%op = fptrunc nnan <4 x double> %inv to <4 x float>
call void @llvm.matrix.column.major.store(<4 x float> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @fpext_2x2(ptr %in, ptr %out) {
; CHECK-LABEL: @fpext_2x2(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[IN:%.*]], align 16
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds float, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x float>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = fpext <2 x float> [[COL_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP2:%.*]] = fpext <2 x float> [[COL_LOAD1]] to <2 x double>
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[OUT:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds double, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x double> [[TMP2]], ptr [[VEC_GEP2]], align 8
; CHECK-NEXT: ret void
;
%inv = load <4 x float>, ptr %in
%op = fpext <4 x float> %inv to <4 x double>
call void @llvm.matrix.column.major.store(<4 x double> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @bitcast_2x2_v4f64_to_v4i64(ptr %in, ptr %out) {
; CHECK-LABEL: @bitcast_2x2_v4f64_to_v4i64(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds double, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = bitcast <2 x double> [[COL_LOAD]] to <2 x i64>
; CHECK-NEXT: [[TMP2:%.*]] = bitcast <2 x double> [[COL_LOAD1]] to <2 x i64>
; CHECK-NEXT: store <2 x i64> [[TMP1]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr inbounds i64, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x i64> [[TMP2]], ptr [[VEC_GEP2]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x double>, ptr %in
%op = bitcast <4 x double> %inv to <4 x i64>
call void @llvm.matrix.column.major.store(<4 x i64> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @bitcast_2x2_v4f64_to_v8i32(ptr %in, ptr %out) {
; CHECK-LABEL: @bitcast_2x2_v4f64_to_v8i32(
; CHECK-NEXT: [[INV:%.*]] = load <4 x double>, ptr [[IN:%.*]], align 32
; CHECK-NEXT: [[OP:%.*]] = bitcast <4 x double> [[INV]] to <8 x i32>
; CHECK-NEXT: [[SPLIT:%.*]] = shufflevector <8 x i32> [[OP]], <8 x i32> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
; CHECK-NEXT: [[SPLIT1:%.*]] = shufflevector <8 x i32> [[OP]], <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
; CHECK-NEXT: store <4 x i32> [[SPLIT]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 4
; CHECK-NEXT: store <4 x i32> [[SPLIT1]], ptr [[VEC_GEP]], align 4
; CHECK-NEXT: ret void
;
%inv = load <4 x double>, ptr %in
%op = bitcast <4 x double> %inv to <8 x i32>
call void @llvm.matrix.column.major.store(<8 x i32> %op, ptr %out, i64 4, i1 false, i32 4, i32 2)
ret void
}
define void @bitcast_2x2_i256_to_v4i64(ptr %in, ptr %out) {
; CHECK-LABEL: @bitcast_2x2_i256_to_v4i64(
; CHECK-NEXT: [[INV:%.*]] = load i256, ptr [[IN:%.*]], align 4
; CHECK-NEXT: [[OP:%.*]] = bitcast i256 [[INV]] to <4 x double>
; CHECK-NEXT: [[SPLIT:%.*]] = shufflevector <4 x double> [[OP]], <4 x double> poison, <2 x i32> <i32 0, i32 1>
; CHECK-NEXT: [[SPLIT1:%.*]] = shufflevector <4 x double> [[OP]], <4 x double> poison, <2 x i32> <i32 2, i32 3>
; CHECK-NEXT: store <2 x double> [[SPLIT]], ptr [[OUT:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds double, ptr [[OUT]], i64 2
; CHECK-NEXT: store <2 x double> [[SPLIT1]], ptr [[VEC_GEP]], align 8
; CHECK-NEXT: ret void
;
%inv = load i256, ptr %in
%op = bitcast i256 %inv to <4 x double>
call void @llvm.matrix.column.major.store(<4 x double> %op, ptr %out, i64 2, i1 false, i32 2, i32 2)
ret void
}
define void @bitcast_2x2_4i64_to_i256(ptr %in, ptr %out) {
; CHECK-LABEL: @bitcast_2x2_4i64_to_i256(
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[IN:%.*]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr inbounds double, ptr [[IN]], i64 2
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = shufflevector <2 x double> [[COL_LOAD]], <2 x double> [[COL_LOAD1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
; CHECK-NEXT: [[OP:%.*]] = bitcast <4 x double> [[TMP1]] to i256
; CHECK-NEXT: store i256 [[OP]], ptr [[OUT:%.*]], align 4
; CHECK-NEXT: ret void
;
%inv = call <4 x double> @llvm.matrix.column.major.load(ptr %in, i64 2, i1 false, i32 2, i32 2)
%op = bitcast <4 x double> %inv to i256
store i256 %op, ptr %out
ret void
}