60b6c53f25
fixes #166206 - Add swizzle support if row index is constant - Add test cases - Add new AST type - Add new LValue for Matrix Row Type - TODO: Make the new LValue a dynamic index version of ExtVectorElt
128 lines
8.9 KiB
HLSL
128 lines
8.9 KiB
HLSL
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
|
|
// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.7-library -disable-llvm-passes -emit-llvm -finclude-default-header -o - %s | FileCheck %s
|
|
|
|
// CHECK-LABEL: define hidden void @_Z9setMatrixRu11matrix_typeILm4ELm4EfEiDv4_f(
|
|
// CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(64) [[M:%.*]], i32 noundef [[INDEX:%.*]], <4 x float> noundef nofpclass(nan inf) [[V:%.*]]) #[[ATTR0:[0-9]+]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK-NEXT: [[V_ADDR:%.*]] = alloca <4 x float>, align 16
|
|
// CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4
|
|
// CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4
|
|
// CHECK-NEXT: store <4 x float> [[V]], ptr [[V_ADDR]], align 16
|
|
// CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[V_ADDR]], align 16
|
|
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3:![0-9]+]], !align [[META4:![0-9]+]]
|
|
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[INDEX_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_LOAD:%.*]] = load <16 x float>, ptr [[TMP1]], align 4
|
|
// CHECK-NEXT: [[TMP3:%.*]] = add i32 0, [[TMP2]]
|
|
// CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x float> [[TMP0]], i32 0
|
|
// CHECK-NEXT: [[TMP5:%.*]] = insertelement <16 x float> [[MATRIX_LOAD]], float [[TMP4]], i32 [[TMP3]]
|
|
// CHECK-NEXT: [[TMP6:%.*]] = add i32 4, [[TMP2]]
|
|
// CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x float> [[TMP0]], i32 1
|
|
// CHECK-NEXT: [[TMP8:%.*]] = insertelement <16 x float> [[TMP5]], float [[TMP7]], i32 [[TMP6]]
|
|
// CHECK-NEXT: [[TMP9:%.*]] = add i32 8, [[TMP2]]
|
|
// CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x float> [[TMP0]], i32 2
|
|
// CHECK-NEXT: [[TMP11:%.*]] = insertelement <16 x float> [[TMP8]], float [[TMP10]], i32 [[TMP9]]
|
|
// CHECK-NEXT: [[TMP12:%.*]] = add i32 12, [[TMP2]]
|
|
// CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x float> [[TMP0]], i32 3
|
|
// CHECK-NEXT: [[TMP14:%.*]] = insertelement <16 x float> [[TMP11]], float [[TMP13]], i32 [[TMP12]]
|
|
// CHECK-NEXT: store <16 x float> [[TMP14]], ptr [[TMP1]], align 4
|
|
// CHECK-NEXT: ret void
|
|
//
|
|
void setMatrix(out float4x4 M, int index, float4 V) {
|
|
M[index] = V;
|
|
}
|
|
|
|
// CHECK-LABEL: define hidden void @_Z15setMatrixScalarRu11matrix_typeILm2ELm1EfEif(
|
|
// CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(8) [[M:%.*]], i32 noundef [[INDEX:%.*]], float noundef nofpclass(nan inf) [[S:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK-NEXT: [[INDEX_ADDR:%.*]] = alloca i32, align 4
|
|
// CHECK-NEXT: [[S_ADDR:%.*]] = alloca float, align 4
|
|
// CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4
|
|
// CHECK-NEXT: store i32 [[INDEX]], ptr [[INDEX_ADDR]], align 4
|
|
// CHECK-NEXT: store float [[S]], ptr [[S_ADDR]], align 4
|
|
// CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[S_ADDR]], align 4
|
|
// CHECK-NEXT: [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP0]], i64 0
|
|
// CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
|
|
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3]], !align [[META4]]
|
|
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[INDEX_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_LOAD:%.*]] = load <2 x float>, ptr [[TMP1]], align 4
|
|
// CHECK-NEXT: [[TMP3:%.*]] = add i32 0, [[TMP2]]
|
|
// CHECK-NEXT: [[TMP4:%.*]] = extractelement <1 x float> [[SPLAT_SPLAT]], i32 0
|
|
// CHECK-NEXT: [[TMP5:%.*]] = insertelement <2 x float> [[MATRIX_LOAD]], float [[TMP4]], i32 [[TMP3]]
|
|
// CHECK-NEXT: store <2 x float> [[TMP5]], ptr [[TMP1]], align 4
|
|
// CHECK-NEXT: ret void
|
|
//
|
|
void setMatrixScalar(out float2x1 M, int index, float S) {
|
|
M[index] = S;
|
|
}
|
|
|
|
// CHECK-LABEL: define hidden void @_Z19setMatrixConstIndexRu11matrix_typeILm4ELm4EiES_(
|
|
// CHECK-SAME: ptr noalias noundef nonnull align 4 dereferenceable(64) [[M:%.*]], <16 x i32> noundef [[N:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[M_ADDR:%.*]] = alloca ptr, align 4
|
|
// CHECK-NEXT: [[N_ADDR:%.*]] = alloca [16 x i32], align 4
|
|
// CHECK-NEXT: store ptr [[M]], ptr [[M_ADDR]], align 4
|
|
// CHECK-NEXT: store <16 x i32> [[N]], ptr [[N_ADDR]], align 4
|
|
// CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr [[N_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_ELEM:%.*]] = extractelement <16 x i32> [[TMP0]], i32 3
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS:%.*]] = insertelement <4 x i32> poison, i32 [[MATRIX_ELEM]], i32 0
|
|
// CHECK-NEXT: [[MATRIX_ELEM1:%.*]] = extractelement <16 x i32> [[TMP0]], i32 7
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS2:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS]], i32 [[MATRIX_ELEM1]], i32 1
|
|
// CHECK-NEXT: [[MATRIX_ELEM3:%.*]] = extractelement <16 x i32> [[TMP0]], i32 11
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS4:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS2]], i32 [[MATRIX_ELEM3]], i32 2
|
|
// CHECK-NEXT: [[MATRIX_ELEM5:%.*]] = extractelement <16 x i32> [[TMP0]], i32 15
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS6:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS4]], i32 [[MATRIX_ELEM5]], i32 3
|
|
// CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3]], !align [[META4]]
|
|
// CHECK-NEXT: store <4 x i32> [[MATRIX_ROW_INS6]], ptr [[TMP1]], align 4
|
|
// CHECK-NEXT: [[TMP2:%.*]] = load <16 x i32>, ptr [[N_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_ELEM7:%.*]] = extractelement <16 x i32> [[TMP2]], i32 2
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS8:%.*]] = insertelement <4 x i32> poison, i32 [[MATRIX_ELEM7]], i32 0
|
|
// CHECK-NEXT: [[MATRIX_ELEM9:%.*]] = extractelement <16 x i32> [[TMP2]], i32 6
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS10:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS8]], i32 [[MATRIX_ELEM9]], i32 1
|
|
// CHECK-NEXT: [[MATRIX_ELEM11:%.*]] = extractelement <16 x i32> [[TMP2]], i32 10
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS12:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS10]], i32 [[MATRIX_ELEM11]], i32 2
|
|
// CHECK-NEXT: [[MATRIX_ELEM13:%.*]] = extractelement <16 x i32> [[TMP2]], i32 14
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS14:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS12]], i32 [[MATRIX_ELEM13]], i32 3
|
|
// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3]], !align [[META4]]
|
|
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr <16 x i32>, ptr [[TMP3]], i32 0, i32 4
|
|
// CHECK-NEXT: store <4 x i32> [[MATRIX_ROW_INS14]], ptr [[TMP4]], align 4
|
|
// CHECK-NEXT: [[TMP5:%.*]] = load <16 x i32>, ptr [[N_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_ELEM15:%.*]] = extractelement <16 x i32> [[TMP5]], i32 1
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS16:%.*]] = insertelement <4 x i32> poison, i32 [[MATRIX_ELEM15]], i32 0
|
|
// CHECK-NEXT: [[MATRIX_ELEM17:%.*]] = extractelement <16 x i32> [[TMP5]], i32 5
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS18:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS16]], i32 [[MATRIX_ELEM17]], i32 1
|
|
// CHECK-NEXT: [[MATRIX_ELEM19:%.*]] = extractelement <16 x i32> [[TMP5]], i32 9
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS20:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS18]], i32 [[MATRIX_ELEM19]], i32 2
|
|
// CHECK-NEXT: [[MATRIX_ELEM21:%.*]] = extractelement <16 x i32> [[TMP5]], i32 13
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS22:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS20]], i32 [[MATRIX_ELEM21]], i32 3
|
|
// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3]], !align [[META4]]
|
|
// CHECK-NEXT: [[TMP7:%.*]] = getelementptr <16 x i32>, ptr [[TMP6]], i32 0, i32 8
|
|
// CHECK-NEXT: store <4 x i32> [[MATRIX_ROW_INS22]], ptr [[TMP7]], align 4
|
|
// CHECK-NEXT: [[TMP8:%.*]] = load <16 x i32>, ptr [[N_ADDR]], align 4
|
|
// CHECK-NEXT: [[MATRIX_ELEM23:%.*]] = extractelement <16 x i32> [[TMP8]], i32 0
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS24:%.*]] = insertelement <4 x i32> poison, i32 [[MATRIX_ELEM23]], i32 0
|
|
// CHECK-NEXT: [[MATRIX_ELEM25:%.*]] = extractelement <16 x i32> [[TMP8]], i32 4
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS26:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS24]], i32 [[MATRIX_ELEM25]], i32 1
|
|
// CHECK-NEXT: [[MATRIX_ELEM27:%.*]] = extractelement <16 x i32> [[TMP8]], i32 8
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS28:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS26]], i32 [[MATRIX_ELEM27]], i32 2
|
|
// CHECK-NEXT: [[MATRIX_ELEM29:%.*]] = extractelement <16 x i32> [[TMP8]], i32 12
|
|
// CHECK-NEXT: [[MATRIX_ROW_INS30:%.*]] = insertelement <4 x i32> [[MATRIX_ROW_INS28]], i32 [[MATRIX_ELEM29]], i32 3
|
|
// CHECK-NEXT: [[TMP9:%.*]] = load ptr, ptr [[M_ADDR]], align 4, !nonnull [[META3]], !align [[META4]]
|
|
// CHECK-NEXT: [[TMP10:%.*]] = getelementptr <16 x i32>, ptr [[TMP9]], i32 0, i32 12
|
|
// CHECK-NEXT: store <4 x i32> [[MATRIX_ROW_INS30]], ptr [[TMP10]], align 4
|
|
// CHECK-NEXT: ret void
|
|
//
|
|
void setMatrixConstIndex(out int4x4 M, int4x4 N ) {
|
|
M[0] = N[3];
|
|
M[1] = N[2];
|
|
M[2] = N[1];
|
|
M[3] = N[0];
|
|
}
|
|
|
|
//.
|
|
// CHECK: [[META3]] = !{}
|
|
// CHECK: [[META4]] = !{i64 4}
|
|
//.
|