srcarroll wrote: The current implementation will emit a `tensor.reshape` op if any of the dims of the input are not factorable (require more than one dynamic dim in the expansion). However, I could instead only emit reshapes for the dims that need it, and then a `tensor.expand_shape` on the rest of the dims. Whichever is preferable to the reviewer(s).
https://github.com/llvm/llvm-project/pull/76003 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
