This repository has been archived by the owner on Nov 25, 2024. It is now read-only.
Forward-merge branch-23.12 to branch-24.02 #102
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
replace `optparse` with `argparser` in `pylibwholegraph` Authors: - Chuang Zhu (https://github.com/chuangz0) Approvers: - https://github.com/dongxuy04 - Brad Rees (https://github.com/BradReesWork) URL: rapidsai#61
The package currently has runtime dependencies on librmm and libraft-headers. However, these packages are header-only, so it's only sensible to require them at build time. This PR removes librmm and libraft-headers from the runtime dependencies of libwholegraph. This PR is needed to fix the conda environments in [unified devcontainers](https://github.com/rapidsai/devcontainers). Authors: - Bradley Dice (https://github.com/bdice) Approvers: - Ray Douglass (https://github.com/raydouglass) URL: rapidsai#96
Replace the random number generator (RNG) implemented by wholegraph with RNG provided by RAFT. It is put forward by issue rapidsai#7 and issue rapidsai#23. Authors: - https://github.com/linhu-nv - Brad Rees (https://github.com/BradReesWork) Approvers: - https://github.com/dongxuy04 - Chuang Zhu (https://github.com/chuangz0) URL: rapidsai#79
Support `gather` and `scatter` operations with NVSHMEM when using the `distributed` embedding.
The way to use `nvshmem` is
```
#build command: ./build.sh --enable-nvshmem ....
global_comm, local_comm = wgth.init_torch_env_and_create_wm_comm(
wgth.get_rank(),
wgth.get_world_size(),
wgth.get_local_rank(),
wgth.get_local_size(),
options.distributed_backend_type
)
```
If `distributed_backend_type` is `"nvshmem"`, then the `gather` and `scatter` operation for distributed wholememory created by `global_comm` will be implemented with `nvshmem`.
Distributed embedding with `cache` is not supported when using nvshmem.
Authors:
- Chuang Zhu (https://github.com/chuangz0)
Approvers:
- Seunghwa Kang (https://github.com/seunghwak)
- Brad Rees (https://github.com/BradReesWork)
raydouglass
approved these changes
Nov 27, 2023
bdice
force-pushed
the
branch-24.02-merge-23.12
branch
from
8fcc2b0
to
e90dda8
Compare
BradReesWork
added
improvement
bdice
force-pushed
the
branch-24.02-merge-23.12
branch
2 times, most recently
from
94fd363
to
9853c62
Compare
|
/ok to test |
|
@BradReesWork Just FYI, we do not want to |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Manual forward merge from 23.12 to 24.02. This PR should not be squashed.
Closes #99.