superlu-dist: add 8.1.1, 8.1.0 and 8.0.0 versions#32558
superlu-dist: add 8.1.1, 8.1.0 and 8.0.0 versions#32558alalazo merged 4 commits intospack:developfrom
Conversation
|
@liuyangzhuan When trying to build [email protected] with CUDA, I am getting some errors: spack spec: spack debug report: |
|
@balos1 the following works for me |
tldahlgren
left a comment
There was a problem hiding this comment.
Confirmed new sha256.
@liuyangzhuan I am running into the same error on LLNL's Lassen with [email protected]. I don't see the gcc versions being mixed anywhere in the build logs. I wonder if it possibly related to the error seen in #32505? |
|
@liuyangzhuan With the rebase the CUDA build is working for me now. I am not sure what the problem was. |
|
@tldahlgren Can this be merged even though gitlab-ci is failing? It looks like the failure is from e4s, and specifically Trilinos. The Trilinos versions being built in the pipeline are incompatible with SuperLU_DIST 8.x.x (see trilinos/Trilinos#10673). |
The CI pipeline checks are required and I cannot override them. @sethrj @eugeneswalker Looks like the |
|
I don't think there are any further trilinos updates available; @jwillenbring could comment perhaps? It sounds as though that there needs to at least be some version restrictions for the superlu compatibility when |
🤦 Even better! |
I think the correct thing to do from a Spack standpoint would be to change https://github.com/spack/spack/blob/develop/var/spack/repos/builtin/packages/trilinos/package.py#L423 to depends_on("[email protected]:7.99.99", when="@13.0.1:13.99.99 +superlu-dist")maybe (also, perhaps the spack version syntax I used could be cleaner)? I believe Trilinos 14 will work with superlu-dist 8.x.x but @jwillenbring and @keitat should be the official word on this. |
|
Should be good now that #32889 was merged. I also added 8.1.1 into this as well since it was released after I opened the PR. |
tldahlgren
left a comment
There was a problem hiding this comment.
Re-confirmed the new sha56. Other changes seem reasonable to me.
|
@liuyangzhuan Do you want to revisit this PR before it gets merged? |
No description provided.