[Mlir-commits] [mlir] [MLIR][XeGPU] Add 2D `vector.multi_reduction` optimization (PR #171154)

[llvmlistbot at llvm.org]

github-actions[bot] wrote:

<!--PREMERGE ADVISOR COMMENT: Windows-->
# :window: Windows x64 Test Results

* 3388 tests passed
* 411 tests skipped
* 1 test failed

## Failed Tests
(click on a test name to see its output)

### MLIR
<details>
<summary>MLIR.Dialect/XeGPU/peephole-optimize.mlir</summary>

```
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 1
c:\_work\llvm-project\llvm-project\build\bin\mlir-opt.exe --xevm-attach-target='module=xevm_* chip=pvc'     --xegpu-optimize-peephole --canonicalize --cse --split-input-file C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir | c:\_work\llvm-project\llvm-project\build\bin\filecheck.exe C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir
# executed command: 'c:\_work\llvm-project\llvm-project\build\bin\mlir-opt.exe' '--xevm-attach-target=module=xevm_* chip=pvc' --xegpu-optimize-peephole --canonicalize --cse --split-input-file 'C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir'
# note: command had no output on stdout or stderr
# executed command: 'c:\_work\llvm-project\llvm-project\build\bin\filecheck.exe' 'C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir'
# .---command stderr------------
# | C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir:285:11: error: CHECK: expected string not found in input
# | // CHECK: %[[ACC:.*]] = arith.constant {layout_result_0 = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0, 1]>} 1.000000e+00 : f32
# |           ^
# | <stdin>:196:79: note: scanning from here
# |  gpu.func @vector_reduce_2d(%arg0: memref<4x16xf32>, %arg1: memref<256xf32>) {
# |                                                                               ^
# | <stdin>:204:2: note: possible intended match here
# |  %4 = vector.broadcast %3 {layout_result_0 = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0]>} : f32 to vector<16xf32>
# |  ^
# | 
# | Input file: <stdin>
# | Check file: C:\_work\llvm-project\llvm-project\mlir\test\Dialect\XeGPU\peephole-optimize.mlir
# | 
# | -dump-input=help explains the following input dump.
# | 
# | Input was:
# | <<<<<<
# |              .
# |              .
# |              .
# |            191: } 
# |            192:  
# |            193: // ----- 
# |            194: module { 
# |            195:  gpu.module @xevm_test [#xevm.target<chip = "pvc">] { 
# |            196:  gpu.func @vector_reduce_2d(%arg0: memref<4x16xf32>, %arg1: memref<256xf32>) { 
# | check:285'0                                                                                   X error: no match found
# |            197:  %cst = arith.constant dense<1.000000e+00> : vector<16xf32> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            198:  %cst_0 = arith.constant {layout_result_0 = #xegpu.layout<lane_layout = [16], lane_data = [1]>} dense<true> : vector<16xi1> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            199:  %cst_1 = arith.constant {layout_result_0 = #xegpu.layout<lane_layout = [16], lane_data = [1]>} dense<0> : vector<16xindex> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            200:  %0 = xegpu.create_nd_tdesc %arg0 : memref<4x16xf32> -> !xegpu.tensor_desc<4x16xf32, #xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            201:  %1 = xegpu.load_nd %0[0, 0] : !xegpu.tensor_desc<4x16xf32, #xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>> -> vector<4x16xf32> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            202:  %2 = vector.multi_reduction <add>, %1, %cst {layout_result_0 = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0]>} [0] : vector<4x16xf32> to vector<16xf32> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            203:  %3 = vector.reduction <add>, %2 {layout_result_0 = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0, 1]>} : vector<16xf32> into f32 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            204:  %4 = vector.broadcast %3 {layout_result_0 = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0]>} : f32 to vector<16xf32> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | check:285'1      ?                                                                                                                                                        possible intended match
# |            205:  xegpu.store %4, %arg1[%cst_1], %cst_0 <{layout = #xegpu.slice<#xegpu.layout<lane_layout = [1, 16], lane_data = [1, 1]>, dims = [0]>}> : vector<16xf32>, memref<256xf32>, vector<16xindex>, vector<16xi1> 
# | check:285'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |            206:  gpu.return 
# | check:285'0     ~~~~~~~~~~~~
# |            207:  } 
# | check:285'0     ~~~
# |            208:  } 
# | check:285'0     ~~~
# |            209: } 
# | check:285'0     ~~
# |            210:  
# | check:285'0     ~
# | >>>>>>
# `-----------------------------
# error: command failed with exit status: 1

--

```
</details>

If these failures are unrelated to your changes (for example tests are broken or flaky at HEAD), please open an issue at https://github.com/llvm/llvm-project/issues and add the `infrastructure` label.

https://github.com/llvm/llvm-project/pull/171154