dcbf2c2ca0
Based on this RFC: https://discourse.llvm.org/t/rfc-allow-the-scalarizer-pass-to-scalarize-vectors-returned-in-structs/82306 LLVM intrinsics do not support out params. To get around this limitation implementers will make intrinsics return structs to capture a return type and an out param. This implementation detail should not impact scalarization since these cases should be elementwise operations. ## Three changes are needed. - The CallInst visitor needs to be updated to handle Structs - A new visitor is needed for `ExtractValue` instructions - finsh needs to be update to handle structs so that insert elements are properly propogated. ## Testing changes - Add support for `llvm.frexp` - Add support for `llvm.dx.splitdouble` fixes https://github.com/llvm/llvm-project/issues/111437
46 lines
2.8 KiB
LLVM
46 lines
2.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: opt -passes='function(scalarizer)' -S -mtriple=dxil-pc-shadermodel6.3-library %s | FileCheck %s
|
|
|
|
define void @test_vector_double_split_void(<2 x double> noundef %d) {
|
|
; CHECK-LABEL: define void @test_vector_double_split_void(
|
|
; CHECK-SAME: <2 x double> noundef [[D:%.*]]) {
|
|
; CHECK-NEXT: [[D_I0:%.*]] = extractelement <2 x double> [[D]], i64 0
|
|
; CHECK-NEXT: [[HLSL_ASUINT_I0:%.*]] = call { i32, i32 } @llvm.dx.splitdouble.i32(double [[D_I0]])
|
|
; CHECK-NEXT: [[D_I1:%.*]] = extractelement <2 x double> [[D]], i64 1
|
|
; CHECK-NEXT: [[HLSL_ASUINT_I1:%.*]] = call { i32, i32 } @llvm.dx.splitdouble.i32(double [[D_I1]])
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
%hlsl.asuint = call { <2 x i32>, <2 x i32> } @llvm.dx.splitdouble.v2i32(<2 x double> %d)
|
|
ret void
|
|
}
|
|
|
|
define noundef <3 x i32> @test_vector_double_split(<3 x double> noundef %d) {
|
|
; CHECK-LABEL: define noundef <3 x i32> @test_vector_double_split(
|
|
; CHECK-SAME: <3 x double> noundef [[D:%.*]]) {
|
|
; CHECK-NEXT: [[D_I0:%.*]] = extractelement <3 x double> [[D]], i64 0
|
|
; CHECK-NEXT: [[HLSL_ASUINT_I0:%.*]] = call { i32, i32 } @llvm.dx.splitdouble.i32(double [[D_I0]])
|
|
; CHECK-NEXT: [[D_I1:%.*]] = extractelement <3 x double> [[D]], i64 1
|
|
; CHECK-NEXT: [[HLSL_ASUINT_I1:%.*]] = call { i32, i32 } @llvm.dx.splitdouble.i32(double [[D_I1]])
|
|
; CHECK-NEXT: [[D_I2:%.*]] = extractelement <3 x double> [[D]], i64 2
|
|
; CHECK-NEXT: [[HLSL_ASUINT_I2:%.*]] = call { i32, i32 } @llvm.dx.splitdouble.i32(double [[D_I2]])
|
|
; CHECK-NEXT: [[DOTELEM0:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I0]], 0
|
|
; CHECK-NEXT: [[DOTELEM01:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I1]], 0
|
|
; CHECK-NEXT: [[DOTELEM02:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I2]], 0
|
|
; CHECK-NEXT: [[DOTELEM1:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I0]], 1
|
|
; CHECK-NEXT: [[DOTELEM13:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I1]], 1
|
|
; CHECK-NEXT: [[DOTELEM14:%.*]] = extractvalue { i32, i32 } [[HLSL_ASUINT_I2]], 1
|
|
; CHECK-NEXT: [[DOTI0:%.*]] = add i32 [[DOTELEM0]], [[DOTELEM1]]
|
|
; CHECK-NEXT: [[DOTI1:%.*]] = add i32 [[DOTELEM01]], [[DOTELEM13]]
|
|
; CHECK-NEXT: [[DOTI2:%.*]] = add i32 [[DOTELEM02]], [[DOTELEM14]]
|
|
; CHECK-NEXT: [[DOTUPTO015:%.*]] = insertelement <3 x i32> poison, i32 [[DOTI0]], i64 0
|
|
; CHECK-NEXT: [[DOTUPTO116:%.*]] = insertelement <3 x i32> [[DOTUPTO015]], i32 [[DOTI1]], i64 1
|
|
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <3 x i32> [[DOTUPTO116]], i32 [[DOTI2]], i64 2
|
|
; CHECK-NEXT: ret <3 x i32> [[TMP1]]
|
|
;
|
|
%hlsl.asuint = call { <3 x i32>, <3 x i32> } @llvm.dx.splitdouble.v3i32(<3 x double> %d)
|
|
%1 = extractvalue { <3 x i32>, <3 x i32> } %hlsl.asuint, 0
|
|
%2 = extractvalue { <3 x i32>, <3 x i32> } %hlsl.asuint, 1
|
|
%3 = add <3 x i32> %1, %2
|
|
ret <3 x i32> %3
|
|
}
|