testing: 1.13.0 with new magma 2.6.2 #146
Conversation
…nda-forge-pinning 2022.10.30.22.56.52
…nda-forge-pinning 2022.11.04.03.22.42
…nda-forge-pinning 2022.11.04.17.37.23
|
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
|
One thing we can do is to reduce the cuda arch list and see if it finishes on time. I will also run this locally tomorrow to test for myself... |
| @@ -73,7 +71,7 @@ outputs: | |||
| # GPU requirements | |||
| - cudnn # [cuda_compiler_version != "None"] | |||
| - nccl # [cuda_compiler_version != "None"] | |||
| - magma # [cuda_compiler_version != "None"] | |||
| - magma 2.6.2 # [cuda_compiler_version != "None"] | |||
There was a problem hiding this comment.
why do you have to be so specific.
There was a problem hiding this comment.
Was just testing per the request here: conda-forge/magma-feedstock#20 (comment)
There was a problem hiding this comment.
Also, this is coming from my own channel, as I assume you have already figured
|
@conda-forge-admin please rerender |
…nda-forge-pinning 2022.11.05.21.08.08
|
lets wait for magma to be released, then rebuild. |
@carterbox wanted us to check before releasing the new version of magma, and I think it is safe for pytorch; but magma will likely not be released until a week or so from now according to the discussion in that PR. |
|
It may make sense to merge #144 without the new magma. Then we can rebuild for magma alongside the patch release (it's been the case that pytorch minor releases have a quick patch release within a few weeks of late...) Up to you though, I am in no particular rush 😃 |
|
We aren't as bad as Jax.... I just pushed to main to trigger a rebuild. |
|
Sorry, as bad as arrow. |
What are you saying here? 🤔 |
|
then number of builds to rebuild. ^_^ juts a joke. |
|
arrow having 5 different branches, making it able to bog down azure ^_^ |
All good - sounds like you're gonna like conda-forge/arrow-cpp-feedstock#875; we're now down to 16 jobs (20 once CUDA on aarch/ppc is reinstated), and that's after doubling the build matrix due to openssl 3. And it runs faster (no more restarts 🥳), despite running the full test suite now.
Yeah, I've set |
|
xref #144 (comment) (I think you meant to continue the convo there...) |
|
Great work! These science packages are really hard to compile! |
Checklist
0(if the version changed)conda-smithy(Use the phrase@conda-forge-admin, please rerenderin a comment in this PR for automated rerendering)