[Mlir-commits] [mlir] [MLIR][XeGPU][VectorToXeGPU] Lower vector.load/store/transfer_read/transfer_write to new offsets syntax (PR #162095)

[Jianhui Li]

================
@@ -121,6 +121,39 @@ isValidGatherScatterParams(Type maskTy, VectorType valueTy,
   return success();
 }
 
+// Verify that number of offsets matches either the source rank or the tdesc
----------------
Jianhui-Li wrote:

I think we should simply check that offsets can't have lower rank than tdesc rank. It shoud be fine if it is larger than tdesc rank. 

I am not sure that it is common practice for the op validation to validate itself using information from producer op. I would rather checking this in transformation or lowering passes.  Any opinion from @adam-smnk ?

https://github.com/llvm/llvm-project/pull/162095