Revert "[Workaround] "-lflang_rt.runtime" not ready so use "-lFortranRuntime … (#656)"#949
Conversation
Bumps [rocm-docs-core](https://github.com/ROCm/rocm-docs-core) from 1.12.1 to 1.13.0. - [Release notes](https://github.com/ROCm/rocm-docs-core/releases) - [Changelog](https://github.com/ROCm/rocm-docs-core/blob/develop/CHANGELOG.md) - [Commits](ROCm/rocm-docs-core@v1.12.1...v1.13.0) --- updated-dependencies: - dependency-name: rocm-docs-core dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> [ROCm/hipBLAS commit: efc5429]
|
What happens if we don't add these fortran libs at all? Do we have fortran tests that verify that confirm the flags are required? |
|
This is going to fine for staging, but will cause problems for mainline until a bulk merge from amd-staging -> amd-mainline lands. Would it be better to use find_library() to find the correctly named fortran library instead? |
8f6f492 to
c5516b1
Compare
Yes, you're right. I’ve modified the code to use find_library() to check whether FortranRuntime exists. If it doesn’t, it will move to use flang_rt.runtime. An alternative approach would be to check the Clang version directly and link the corresponding libraries, but using find_library() may offer better flexibility. |
c5516b1 to
ff0370d
Compare
The Fortran libraries are linked due to the use of LAPACK for validation. However, since BLIS is now the default validation method, we may consider removing the LAPACK-based validation. If we proceed with that, the linked Fortran libraries may no longer be necessary. |
|
Linked the correct Fortran library, verified passed. http://rocm-ci.amd.com/job/compute-rocm-dkms-component-staging-compiler/8740/ |
|
I am interested in resolving some outstanding complexity around these fortran runtime libraries, for example, why are we supporting both both flang_rt.runtime and FortranRuntime, are these both really required? Either way, this seems acceptable for now given the following webpages that I reviewed: |
|
Changes to hipsparselt code coverage were made recently. I suggest re-triggering that pipeline. |
* Docs: Add environment variable reference page * Fix [ROCm/rocSOLVER commit: 61182e9]
This reverts commit c6c0a69.
for SWDEV-544106