[Mlir-commits] [mlir] [MLIR][XeGPU] Introduce `xegpu::uArch` usage in target-sensitive passes (PR #163801)

[Jianhui Li]

================
@@ -42,12 +40,61 @@ struct Xe2Plus : public uArch {
               &instrs = {})
       : uArch(archName, archDescription, regInfo, cacheInfo, instrs),
         xeCore(xeCore) {}
+  int getSubgroupSize() const override { return 16; }
+  unsigned getPackedFormatBitSizeGatherScatter() const override { return 32; }
+  unsigned getPackedFormatBitSize() const override { return 16; }
----------------
Jianhui-Li wrote:

For C, the result is f32 so no packing needed. 

https://github.com/llvm/llvm-project/pull/163801