Skip to content

Forward merge branch-24.06 into branch-24.08#4489

Merged
rapids-bot[bot] merged 20 commits intorapidsai:branch-24.08from
nv-rliu:branch-24.08-merge-24.06
Jun 19, 2024
Merged

Forward merge branch-24.06 into branch-24.08#4489
rapids-bot[bot] merged 20 commits intorapidsai:branch-24.08from
nv-rliu:branch-24.08-merge-24.06

Conversation

@nv-rliu
Copy link
Copy Markdown
Contributor

@nv-rliu nv-rliu commented Jun 14, 2024

Replaces #4476

nv-rliu and others added 8 commits June 7, 2024 11:30
…#4475)

Addresses rapidsai#4474 

Currently `openmpi=5.0.3-hfd7b305_105` is blocking our CI `cpp_build`
job.

Most likely introduced by this PR:
conda-forge/openmpi-feedstock#158

This PR will unblock cugraph development until the issues are fixed.
Once that happens, the version pinning should be removed.
…sai#4464)

PyTorch 2.2+ is incompatible with the NCCL version on our containers.
Normally, this would not be an issue, but there is a bug in CuPy that
loads the system NCCL instead of the user NCCL. This PR binds the
PyTorch test dependency version to get around this issue.

---------

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
Co-authored-by: Ralph Liu <137829296+nv-rliu@users.noreply.github.com>
Co-authored-by: James Lamb <jlamb@nvidia.com>
Copy link
Copy Markdown
Member

@alexbarghi-nv alexbarghi-nv left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

@alexbarghi-nv alexbarghi-nv added feature request New feature or request non-breaking Non-breaking change labels Jun 14, 2024
@alexbarghi-nv
Copy link
Copy Markdown
Member

@nv-rliu I think this should be renamed "Forward merge branch-24.06 into branch-24.08

@tingyu66 tingyu66 changed the title Forward merge branch-24.08 into branch-24.06 Forward merge branch-24.06 into branch-24.08 Jun 14, 2024
@bdice bdice mentioned this pull request Jun 17, 2024
Comment thread ci/build_wheel.sh Outdated
Comment on lines +60 to +61
PARALLEL_LEVEL=$(python -c \
"from math import ceil; from multiprocessing import cpu_count; print(ceil(cpu_count()/2))")
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we recently added a PARALLEL_LEVEL environment variable to the rapids-configure-sccache script below:

That script is sourced earlier in this file.

Therefore you could simply use the value defined from that script instead of redefining it here.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Great. Just wanted to be aggressive here to rule out the impact of hyperthreading. Using all 64 logical cores to compile cugraph did fail at times on my workstation (w/ Threadripper 3975WX).

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think hyperthreading is a factor on these CI machines -- we get the "real" number of cores, afaik. We should be safe to remove this. Let's wait to push until after CI runs on the current commit (I'm testing something else at the moment).

Comment thread ci/build_wheel.sh Outdated
@tingyu66 tingyu66 marked this pull request as draft June 18, 2024 12:55
@bdice bdice marked this pull request as ready for review June 18, 2024 23:06
@bdice
Copy link
Copy Markdown
Contributor

bdice commented Jun 19, 2024

I'm going to go ahead and trigger a merge. Builds all succeeded.

@bdice
Copy link
Copy Markdown
Contributor

bdice commented Jun 19, 2024

/merge

@rapids-bot rapids-bot Bot merged commit f519ac1 into rapidsai:branch-24.08 Jun 19, 2024
@nv-rliu nv-rliu deleted the branch-24.08-merge-24.06 branch June 24, 2024 13:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ci conda feature request New feature or request non-breaking Non-breaking change python

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants