[Mlir-commits] [llvm] [mlir] [mlir][EmitC] Add MathToEmitC pass for math function lowering to EmitC (PR #113799)

[Marius Brehler]

https://github.com/marbre requested changes to this pull request.

Thanks for the PR! I think we had a discussion before that this could be nice but seems there was no strong need before.

While I like the approach, I have some concerns regarding non scalar-types. From a quick look it seems
```mlir
%f = math.absf %g : tensor<4xf32>
```
would be converted to valid EmitC and the emitter would translate to C / C++ without an error. I think this needs to be addressed as we cannot in the emitter.

https://github.com/llvm/llvm-project/pull/113799