Dandandan edited a comment on pull request #2124:
URL:
https://github.com/apache/arrow-datafusion/pull/2124#issuecomment-1084366628
> I think we are good as long as we don't invoke rust functions from the
JIT'd code. I am curious if cranelift will be smart enough to auto vectorize
the loop though :)
>
> If we keep the code gen API similar, then when we switch to LLVM, the
migration shouldn't be too hard.
:100: Maybe it could event coexist together (both a Cranelift / LLVM
backend).
Also not sure on the autovectorization, though I am hopeful we might be able
to instrument cranelift enough to have it compile to SIMD instructions, given
that we now more about the data types and code than a generic for-loop.
(there are for example SIMD types
https://docs.rs/cranelift/latest/cranelift/prelude/types/index.html).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
