[Mlir-commits] [mlir] [mlir][vector] Add verification for incorrect vector.extract (PR #115824)

[Jakub Kuderski]

kuhar wrote:

> This makes me think that the reducing the number of possible representations of scalars as vectors could be a good thing for SPIR-V?

Reducing the number of representations for 1-element vectors won't help -- all the SPIR-V conversions check for is whether something became a scalar or not.

When it comes to eliminating 1-element vectors all together, we could make the conversion happen somewhere higher up the stack (e.g., when going to arith/vector), but we'd most likely run into the same issues higher up the stack. This is under the assumption that higher-level input IR allows for something equivalent (e.g., 1-element tensors).

https://github.com/llvm/llvm-project/pull/115824