[Mlir-commits] [mlir] [mlir][gpu] reverse parallel loop to gpu dimension mapping order. (PR #79592)

Jungwook Park llvmlistbot at llvm.org
Fri Jan 26 04:53:49 PST 2024


jungpark-mlir wrote:

Also thought about further optimizations but it could be better to be a separate pass.
We can add an option to choose the order but this simply looks more sensible to me now, feel free to discuss.

https://github.com/llvm/llvm-project/pull/79592


More information about the Mlir-commits mailing list