jhuber6 wrote:
> I think we could all benefit from some documentation describing how the SYCL
> compiler flow is intended to work, what tools are added/modified, and what
> the expected outputs are at each compiler phase. Without some idea of the
> architecture of what is being built changes like this and #110771 don't make
> sense and are really difficult to review.
>
> This isn't like introducing a new backend or something where we have a
> defined and understood architecture to evaluate against. We're getting
> patches that only represent small parts of the whole without any context to
> understand how it will all come together in the end.
>
> What worries me is that from the parts I see, it doesn't align with any of
> the community design principles.
>From what I understand, they want to integrate SYCL linking into the existing
>linker wrapper framework. That tool basically extracts embedded offload
>binaries and then calls `clang --target=${target} extracted.o -o image` and
>then embeds that into the runtime. They wanted to know how to do that so my
>suggestion was to make something that looked like a linker so that control
>flow worked. I'm not familiar with the other serialization format however.
https://github.com/llvm/llvm-project/pull/112245
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
