shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity

[MLIR] Add lowerings for atan and atan2 to ROCDL intrinsics

Browse Source 
Differential Revision: https://reviews.llvm.org/D93123
This commit is contained in:
Frederik Gossen 2020-12-11 18:00:17 +01:00
parent 617cd01a4b
commit 1c6bc2c0b5
2 changed files with 34 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
mlir/lib/Conversion/GPUToROCDL/LowerGpuOpsToROCDLOps.cpp
View File

@ -105,6 +105,10 @@ void mlir::populateGpuToROCDLConversionPatterns(
GPUFuncOpLowering<5>, GPUReturnOpLowering>(converter);
patterns.insert<OpToFuncCallLowering<AbsFOp>>(converter, "__ocml_fabs_f32",
"__ocml_fabs_f64");
patterns.insert<OpToFuncCallLowering<AtanOp>>(converter, "__ocml_atan_f32",
"__ocml_atan_f64");
patterns.insert<OpToFuncCallLowering<Atan2Op>>(converter, "__ocml_atan2_f32",
"__ocml_atan2_f64");
patterns.insert<OpToFuncCallLowering<CeilFOp>>(converter, "__ocml_ceil_f32",
"__ocml_ceil_f64");
patterns.insert<OpToFuncCallLowering<CosOp>>(converter, "__ocml_cos_f32",

30
mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir
View File

@ -280,3 +280,33 @@ gpu.module @test_module {
std.return %result32, %result64 : f32, f64
}
}
// -----
gpu.module @test_module {
// CHECK: llvm.func @__ocml_atan_f32(!llvm.float) -> !llvm.float
// CHECK: llvm.func @__ocml_atan_f64(!llvm.double) -> !llvm.double
// CHECK-LABEL: func @gpu_atan
func @gpu_atan(%arg_f32 : f32, %arg_f64 : f64) -> (f32, f64) {
%result32 = std.atan %arg_f32 : f32
// CHECK: llvm.call @__ocml_atan_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
%result64 = std.atan %arg_f64 : f64
// CHECK: llvm.call @__ocml_atan_f64(%{{.*}}) : (!llvm.double) -> !llvm.double
std.return %result32, %result64 : f32, f64
}
}
// -----
gpu.module @test_module {
// CHECK: llvm.func @__ocml_atan2_f32(!llvm.float, !llvm.float) -> !llvm.float
// CHECK: llvm.func @__ocml_atan2_f64(!llvm.double, !llvm.double) -> !llvm.double
// CHECK-LABEL: func @gpu_atan2
func @gpu_atan2(%arg_f32 : f32, %arg_f64 : f64) -> (f32, f64) {
%result32 = std.atan2 %arg_f32, %arg_f32 : f32
// CHECK: llvm.call @__ocml_atan2_f32(%{{.*}}) : (!llvm.float, !llvm.float) -> !llvm.float
%result64 = std.atan2 %arg_f64, %arg_f64 : f64
// CHECK: llvm.call @__ocml_atan2_f64(%{{.*}}) : (!llvm.double, !llvm.double) -> !llvm.double
std.return %result32, %result64 : f32, f64
}
}