[Mlir-commits] [mlir] [MLIR][XeGPU] Improve deinterleave/interleave/dpas_mx ops handling (PR #197223)

llvmlistbot at llvm.org llvmlistbot at llvm.org
Tue May 12 08:51:10 PDT 2026


github-actions[bot] wrote:

<!--PREMERGE ADVISOR COMMENT: Linux-->
# :penguin: Linux x64 Test Results

* 8031 tests passed
* 615 tests skipped
* 1 test failed

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

### MLIR
<details>
<summary>MLIR.Dialect/XeGPU/sg-to-wi-experimental-unit.mlir</summary>

```
Exit Code: 1

Command Output (stdout):
--
# RUN: at line 2
/home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/mlir-opt  --xevm-attach-target='module=xevm_* chip=cri' --allow-unregistered-dialect  --test-xegpu-sg-to-wi-distribute-experimental --split-input-file /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir | /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/FileCheck /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir
# executed command: /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/mlir-opt '--xevm-attach-target=module=xevm_* chip=cri' --allow-unregistered-dialect --test-xegpu-sg-to-wi-distribute-experimental --split-input-file /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir
# note: command had no output on stdout or stderr
# executed command: /home/gha/actions-runner/_work/llvm-project/llvm-project/build/bin/FileCheck /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir
# .---command stderr------------
# | /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir:1297:11: error: CHECK: expected string not found in input
# | // CHECK: %[[SCALE_A:.*]] = vector.shape_cast %[[CST_1]] : vector<8x1xf8E8M0FNU> to vector<8xf8E8M0FNU>
# |           ^
# | <stdin>:956:78: note: scanning from here
# |  %3 = vector.shape_cast %cst_2 : vector<2x1xf8E8M0FNU> to vector<2xf8E8M0FNU>
# |                                                                              ^
# | <stdin>:956:78: note: with "CST_1" equal to "cst_1"
# |  %3 = vector.shape_cast %cst_2 : vector<2x1xf8E8M0FNU> to vector<2xf8E8M0FNU>
# |                                                                              ^
# | <stdin>:958:2: note: possible intended match here
# |  %5 = vector.shape_cast %4 : vector<8xf32> to vector<8x1xf32>
# |  ^
# | 
# | Input file: <stdin>
# | Check file: /home/gha/actions-runner/_work/llvm-project/llvm-project/mlir/test/Dialect/XeGPU/sg-to-wi-experimental-unit.mlir
# | 
# | -dump-input=help explains the following input dump.
# | 
# | Input was:
# | <<<<<<
# |               .
# |               .
# |               .
# |             951:  %cst_1 = arith.constant dense<1.000000e+00> : vector<8x1xf8E8M0FNU> 
# |             952:  %cst_2 = arith.constant dense<5.000000e-01> : vector<2x1xf8E8M0FNU> 
# |             953:  %0 = vector.shape_cast %cst : vector<8x1xf8E4M3FN> to vector<8xf8E4M3FN> 
# |             954:  %1 = vector.shape_cast %cst_0 : vector<16x1xf8E4M3FN> to vector<16xf8E4M3FN> 
# |             955:  %2 = vector.shape_cast %cst_1 : vector<8x1xf8E8M0FNU> to vector<8xf8E8M0FNU> 
# |             956:  %3 = vector.shape_cast %cst_2 : vector<2x1xf8E8M0FNU> to vector<2xf8E8M0FNU> 
# | check:1297'0                                                                                  X error: no match found
# | check:1297'1                                                                                    with "CST_1" equal to "cst_1"
# |             957:  %4 = xegpu.dpas_mx %0, %1 scale_a = %2 scale_b = %3 : (vector<8xf8E4M3FN>, vector<16xf8E4M3FN>, vector<8xf8E8M0FNU>, vector<2xf8E8M0FNU>) -> vector<8xf32> 
# | check:1297'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# |             958:  %5 = vector.shape_cast %4 : vector<8xf32> to vector<8x1xf32> 
# | check:1297'0     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# | check:1297'2      ?                                                             possible intended match
# |             959:  gpu.return 
# | check:1297'0     ~~~~~~~~~~~~
# |             960:  } 
# | check:1297'0     ~~~
# |             961:  } 
# | check:1297'0     ~~~
# |             962: } 
# | check:1297'0     ~~
# |             963:  
# | check:1297'0     ~
# | >>>>>>
# `-----------------------------
# 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/197223


More information about the Mlir-commits mailing list