Skip to content

Conversation

jhuber6
Copy link
Contributor

@jhuber6 jhuber6 commented Feb 11, 2025

Summary:
We do not need to build for all of these architectures. Since the
infrastructure uses the natively created packager binary we can rely on
the support for generic targets. This is the same thing the OpenMP /
libc projects use now.

Depends on #126655

@llvmbot llvmbot added the flang Flang issues not falling into any other category label Feb 11, 2025
Summary:
We do not need to build for all of these architectures. Since the
infrastructure uses the natively created packager binary we can rely on
the support for generic targets. This is the same thing the OpenMP /
libc projects use now.

Depends on llvm#126655
@Meinersbur Meinersbur changed the title [Flang] Stop building multiple verisons of the OpenMP library [Flang] Stop building multiple versions of the OpenMP library Feb 11, 2025
@vzakhari
Copy link
Contributor

Can you please clarify what generic architecture means in this case? What device image we will have in the end?

@jhuber6
Copy link
Contributor Author

jhuber6 commented Feb 12, 2025

Can you please clarify what generic architecture means in this case? What device image we will have in the end?

Same as this https://godbolt.org/z/daTsz35jv, just doesn't set target-cpu attributes in the IR. The flang runtime probably doesn't make any use of gfx specific builtins so it doesn't make much of a difference.

@jhuber6
Copy link
Contributor Author

jhuber6 commented May 6, 2025

Closing in favor of using the -DRUNTIMES_amdgcn-amd-amdhsa_LLVM_ENABLE_RUNTIMES=amdgcn-amd-amdhsa version.

@jhuber6 jhuber6 closed this May 6, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
flang Flang issues not falling into any other category
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants