Skip to content

[SYCL][ROCm] Build lld when ROCm for AMD is enabled #4402

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 26, 2021

Conversation

npmiller
Copy link
Contributor

This patch ensures lld is built by default when the pi_rocm plugin for AMD is enabled as lld is required for linking with that setup.

This should address: #4390

@bader
Copy link
Contributor

bader commented Aug 25, 2021

I see that we have at least to options to get lld:

  1. Find it in the system - [Buildbot][NFC] Add option to use LLD as linker #3866.
  2. We have lld sources, so we can build them.

Which of these two we should prefer? Do we need to support them both? If we build lld from sources, can be drop #3866?

@alexbatashev
Copy link
Contributor

I see that we have at least to options to get lld:

  1. Find it in the system - [Buildbot][NFC] Add option to use LLD as linker #3866.
  2. We have lld sources, so we can build them.

Which of these two we should prefer? Do we need to support them both? If we build lld from sources, can be drop #3866?

I don’t mind always building lld (we may even enable it in CI, I think), but I’m not sure how reliable multistage build is, never tried it before.

@npmiller
Copy link
Contributor Author

I see that we have at least to options to get lld:

1. Find it in the system - [[Buildbot][NFC] Add option to use LLD as linker #3866](https://github.com/intel/llvm/pull/3866).

2. We have `lld` sources, so we can build them.

Which of these two we should prefer? Do we need to support them both? If we build lld from sources, can be drop #3866?

I think #3866 is still useful even if we always build lld because we may not always want to use lld as the linker for the rest of the build.

I suppose we could skip building lld if --use-lld is passed as it should be available on the system, but lld is also a fairly small part of the build so I think it's fine to have both options.

@bader bader linked an issue Aug 26, 2021 that may be closed by this pull request
@bader bader added the hip Issues related to execution on HIP backend. label Aug 26, 2021
@bader bader merged commit 1937dde into intel:sycl Aug 26, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
hip Issues related to execution on HIP backend.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[ROCm] Problem compiling for AMD GPUs
4 participants