[Mlir-commits] [mlir] [mlir][xegpu] Raise the reduction vector size to match maxReduceVectorSize (PR #225122)

[Jianhui Li]

================
@@ -2436,7 +2540,7 @@ xegpu::SliceAttr xegpu::setupMultiReductionResultLayout(
   auto context = srcVecTy.getContext();
 
   const int subgroupSize = uArch->getSubgroupSize();
-  int64_t maxReduceVectorSize = 1; // could extend to spirv vector Size
+  int64_t maxReduceVectorSize = 16; // extend to spirv vector Size
----------------
Jianhui-Li wrote:

Unfortunately it is not a HW size. It is the max vector size of SPIRV IR under vector 16 capability (a common setting in OpenCL compute environment).  Not sure how to get this number from anywhere. 

https://github.com/llvm/llvm-project/pull/225122