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

[Artem Kroviakov]

================
@@ -565,10 +567,10 @@ TensorDescType::verify(llvm::function_ref<InFlightDiagnostic()> emitError,
 
   // for gather and scatter ops, Low-precision types are packed in 32-bit units.
   unsigned bitWidth = elementType.getIntOrFloatBitWidth();
-  int chunkAlignmentFactor =
-      bitWidth < targetinfo::packedSizeInBitsForGatherScatter
-          ? targetinfo::packedSizeInBitsForGatherScatter / bitWidth
-          : 1;
+  constexpr int packingBitSizeGatherScatter{32};
----------------
akroviakov wrote:

Moved it to be a constexpr constant under `xegpu::uArch` namespace.

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