751a546fa9
Issue: https://github.com/llvm/llvm-project/issues/99172 - [x] Implement `WavePrefixSum` clang builtin - [x] Link `WavePrefixSum` clang builtin with `hlsl_intrinsics.h` - [x] Add sema checks for `WavePrefixSum` to `CheckHLSLBuiltinFunctionCall` in `SemaChecking.cpp` - [x] Add codegen for `WavePrefixSum` to `EmitHLSLBuiltinExpr` in `CGBuiltin.cpp` - [x] Add codegen tests to `clang/test/CodeGenHLSL/builtins/WavePrefixSum.hlsl` - [x] Add sema tests to `clang/test/SemaHLSL/BuiltIns/WavePrefixSum-errors.hlsl` - [x] Create the `int_dx_WavePrefixSum` intrinsic in `IntrinsicsDirectX.td` - [x] Create the `DXILOpMapping` of `int_dx_WavePrefixSum` to `121` in `DXIL.td` - [x] Create the `WavePrefixSum.ll` and `WavePrefixSum_errors.ll` tests in `llvm/test/CodeGen/DirectX/` - [x] Create the `int_spv_WavePrefixSum` intrinsic in `IntrinsicsSPIRV.td` - [x] In SPIRVInstructionSelector.cpp create the `WavePrefixSum` lowering and map it to `int_spv_WavePrefixSum` in `SPIRVInstructionSelector::selectIntrinsic`. - [x] Create SPIR-V backend test case in `llvm/test/CodeGen/SPIRV/hlsl-intrinsics/WavePrefixSum.ll` I also added a new macro `GENERATE_HLSL_INTRINSIC_FUNCTION_SELECT_UNSIGNED` in conjunction with the new function `getUnsignedIntrinsicVariant` to make selecting unsigned variants of the intrinsic easier. As a result, I was able to replace `getWaveActiveSumIntrinsic`, `getWaveActiveMaxIntrinsic`, and `getWaveActiveMinIntrinsic` using the new macro.
128 lines
3.4 KiB
LLVM
128 lines
3.4 KiB
LLVM
; RUN: opt -S --passes="print-dx-shader-flags" 2>&1 %s | FileCheck %s
|
|
;
|
|
; Test that we have the correct shader flags to indicate that there are wave
|
|
; ops set at the module level
|
|
;
|
|
; CHECK: ; Shader Flags Value: [[WAVE_FLAG:0x00080000]]
|
|
; CHECK: ; Note: shader requires additional functionality:
|
|
; CHECK-NEXT: ; Wave level operations
|
|
; CHECK-NEXT: ; Note: extra DXIL module flags:
|
|
|
|
target triple = "dxil-pc-shadermodel6.7-library"
|
|
%dx.types.fouri32 = type { i32, i32, i32, i32 }
|
|
|
|
; Test the indiviual ops that they have the same Shader Wave flag at the
|
|
; function level to ensure that each op is setting it accordingly
|
|
|
|
define noundef i1 @wave_is_first_lane() {
|
|
entry:
|
|
; CHECK: Function wave_is_first_lane : [[WAVE_FLAG]]
|
|
%ret = call i1 @llvm.dx.wave.is.first.lane()
|
|
ret i1 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_getlaneindex() {
|
|
entry:
|
|
; CHECK: Function wave_getlaneindex : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.getlaneindex()
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i1 @wave_any(i1 %x) {
|
|
entry:
|
|
; CHECK: Function wave_any : [[WAVE_FLAG]]
|
|
%ret = call i1 @llvm.dx.wave.any(i1 %x)
|
|
ret i1 %ret
|
|
}
|
|
|
|
define noundef i1 @wave_all(i1 %x) {
|
|
entry:
|
|
; CHECK: Function wave_all : [[WAVE_FLAG]]
|
|
%ret = call i1 @llvm.dx.wave.all(i1 %x)
|
|
ret i1 %ret
|
|
}
|
|
|
|
define noundef i1 @wave_readlane(i1 %x, i32 %idx) {
|
|
entry:
|
|
; CHECK: Function wave_readlane : [[WAVE_FLAG]]
|
|
%ret = call i1 @llvm.dx.wave.readlane.i1(i1 %x, i32 %idx)
|
|
ret i1 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_sum(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_sum : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.sum.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_usum(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_usum : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.usum.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_max(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_max : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.max.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_umax(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_umax : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.umax.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_min(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_min : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.min.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_reduce_umin(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_reduce_umin : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.reduce.umin.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define void @wave_active_countbits(i1 %expr) {
|
|
entry:
|
|
; CHECK: Function wave_active_countbits : [[WAVE_FLAG]]
|
|
%0 = call i32 @llvm.dx.wave.active.countbits(i1 %expr)
|
|
ret void
|
|
}
|
|
|
|
define void @wave_active_ballot(i1 %expr) {
|
|
entry:
|
|
; CHECK: Function wave_active_ballot : [[WAVE_FLAG]]
|
|
%0 = call %dx.types.fouri32 @llvm.dx.wave.ballot(i1 %expr)
|
|
ret void
|
|
}
|
|
|
|
define void @wave_prefix_bit_count(i1 %expr) {
|
|
entry:
|
|
; CHECK: Function wave_prefix_bit_count : [[WAVE_FLAG]]
|
|
%0 = call i32 @llvm.dx.wave.prefix.bit.count(i1 %expr)
|
|
ret void
|
|
}
|
|
|
|
define noundef i32 @wave_prefix_sum(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_prefix_sum : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.prefix.sum.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|
|
|
|
define noundef i32 @wave_prefix_usum(i32 noundef %x) {
|
|
entry:
|
|
; CHECK: Function wave_prefix_usum : [[WAVE_FLAG]]
|
|
%ret = call i32 @llvm.dx.wave.prefix.usum.i32(i32 %x)
|
|
ret i32 %ret
|
|
}
|