[Mlir-commits] [llvm] [mlir] [XeGPU][Transform] Add XeGPU array length optimization pass (PR #194062)

[Jianhui Li]

================
@@ -79,6 +79,8 @@ void buildGPUPassPipeline(OpPassManager &pm,
     pm.addNestedPass<gpu::GPUModuleOp>(
         xegpu::createXeGPUPropagateLayout(instDataOptions));
     pm.addNestedPass<gpu::GPUModuleOp>(xegpu::createXeGPUBlocking());
+    pm.addNestedPass<gpu::GPUModuleOp>(
----------------
Jianhui-Li wrote:

I think we should remove this now since it is now merged to peephole optimization. 

https://github.com/llvm/llvm-project/pull/194062