[mlir] Added Abs, Ceil and Cos to GPU lowering conversion.

[ftynse]

Looks good to me in general, please see inline.

mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir

48

What's the point of exercising the same operation twice?

[pifon2a]

mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir

48

I think this is copied from expf test where I was checking that the symbols are not added twice to the symbol table. I think here it is enough to exercise the op only once.

[merge_guards_bot]

Unit tests: pass. 61326 tests passed, 0 failed and 737 were skipped.

clang-tidy: fail. Please fix clang-tidy findings.

clang-format: fail. Please format your changes with clang-format by running git-clang-format HEAD^ or applying this patch.

Build artifacts: diff.json, clang-tidy.txt, clang-format.patch, CMakeCache.txt, console-log.txt, test-results.xml

[dfki-jugr]

Update clang-format issues.

[dfki-jugr]

Removed unnecessary test cases in test files.

[pifon2a]

Please, reformat the code and submit.

[rriddle]

When uploading a diff, can you make sure to diff against HEAD instead of the previous commit? I couldn't see any of your changes and had to manually look through the diff history.

mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp

715 ↗

(On Diff #236981)

Can we merge all of these instead?

addIllegalOp<..., ..., ..., ...>();

mlir/lib/Conversion/GPUToROCDL/LowerGpuOpsToROCDLOps.cpp

66 ↗

(On Diff #236981)

Same here.

[dfki-jugr]

Merged template arguments in AddIllegalOP.

[ftynse]

Do you guys have commit access or should I land this for you?

[pifon2a]

In D72439#1816470, @ftynse wrote:

Do you guys have commit access or should I land this for you?

I have commit access, I can land it.

mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir

 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @__nv_fabsf(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @__nv_fabs(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_fabs
   func @gpu_fabs(%arg_f32 : f32, %arg_f64 : f64) {
-    %fabs_f32 = std.absf %arg_f32 : f32
-    // CHECK: llvm.call @__nv_fabsf(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.absf %fabs_f32 : f32
+    %result32 = std.absf %arg_f32 : f32
     // CHECK: llvm.call @__nv_fabsf(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.absf %arg_f64 : f64
     // CHECK: llvm.call @__nv_fabs(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @__nv_ceilf(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @__nv_ceil(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_ceil
   func @gpu_ceil(%arg_f32 : f32, %arg_f64 : f64) {
-    %ceil_f32 = std.ceilf %arg_f32 : f32
-    // CHECK: llvm.call @__nv_ceilf(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.ceilf %ceil_f32 : f32
+    %result32 = std.ceilf %arg_f32 : f32
     // CHECK: llvm.call @__nv_ceilf(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.ceilf %arg_f64 : f64
     // CHECK: llvm.call @__nv_ceil(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @__nv_cosf(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @__nv_cos(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_cos
   func @gpu_cos(%arg_f32 : f32, %arg_f64 : f64) {
-    %cos_f32 = std.cos %arg_f32 : f32
-    // CHECK: llvm.call @__nv_cosf(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.cos %cos_f32 : f32
+    %result32 = std.cos %arg_f32 : f32
     // CHECK: llvm.call @__nv_cosf(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.cos %arg_f64 : f64
     // CHECK: llvm.call @__nv_cos(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----

mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir

 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @_ocml_fabs_f32(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @_ocml_fabs_f64(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_fabs
   func @gpu_fabs(%arg_f32 : f32, %arg_f64 : f64) {
-    %fabs_f32 = std.absf %arg_f32 : f32
-    // CHECK: llvm.call @_ocml_fabs_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.absf %fabs_f32 : f32
+    %result32 = std.absf %arg_f32 : f32
     // CHECK: llvm.call @_ocml_fabs_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.absf %arg_f64 : f64
     // CHECK: llvm.call @_ocml_fabs_f64(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @_ocml_ceil_f32(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @_ocml_ceil_f64(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_ceil
   func @gpu_ceil(%arg_f32 : f32, %arg_f64 : f64) {
-    %ceil_f32 = std.ceilf %arg_f32 : f32
-    // CHECK: llvm.call @_ocml_ceil_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.ceilf %ceil_f32 : f32
+    %result32 = std.ceilf %arg_f32 : f32
     // CHECK: llvm.call @_ocml_ceil_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.ceilf %arg_f64 : f64
     // CHECK: llvm.call @_ocml_ceil_f64(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----
 
 module attributes {gpu.kernel_module} {
   // CHECK: llvm.func @_ocml_cos_f32(!llvm.float) -> !llvm.float
   // CHECK: llvm.func @_ocml_cos_f64(!llvm.double) -> !llvm.double
   // CHECK-LABEL: func @gpu_cos
   func @gpu_cos(%arg_f32 : f32, %arg_f64 : f64) {
-    %cos_f32 = std.cos %arg_f32 : f32
-    // CHECK: llvm.call @_ocml_cos_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
-    %result_f32 = std.cos %cos_f32 : f32
+    %result32 = std.cos %arg_f32 : f32
     // CHECK: llvm.call @_ocml_cos_f32(%{{.*}}) : (!llvm.float) -> !llvm.float
     %result64 = std.cos %arg_f64 : f64
     // CHECK: llvm.call @_ocml_cos_f64(%{{.*}}) : (!llvm.double) -> !llvm.double
     std.return
   }
 }
 
 // -----