alamb commented on PR #2587: URL: https://github.com/apache/arrow-datafusion/pull/2587#issuecomment-1135017030
> It just looks like a bare translation of what we build. So I suspect the vectorization is not done here (after translation). Further, I find this [I64X8 type](https://docs.rs/cranelift-codegen/latest/cranelift_codegen/ir/types/constant.I64X8.html) from the document (we are currently using [I64](https://docs.rs/cranelift-codegen/latest/cranelift_codegen/ir/types/constant.I64.html)). Perhaps this means that we need to manually vectorize our computation. I believe this is a known limitation with cranelift -- we can also potentially consider using llvm in the future, but that would likely result in longer query planning times. I believe @tustvold has thought about this as well. Writing some sort of basic vectorization optimizations in cranelift (or in datafusion) is also a possbility -- 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]
