Allow disabling RMM in benchmarks #1352
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
Allows disabling RMM in benchmarks via a new option `--disable-rmm`. This change makes benchmarks a little more similar to RMM setup in `LocalCUDACluster`/`dask cuda worker`, where not specifying `rmm-pool-size` or specifying `None` as its value entirely disables setting up RMM as the default allocator. Since for benchmarks it's desired that the default is having an RMM pool we cannot change the default `--rmm-pool-size` to `None` as that would make benchmarks run much slower by default, therefore `--disable-rmm` is the closest we can make this to the rest of Dask-CUDA. Additionally add `--rmm-maximum-pool-size` for benchmarks.
pentschev
added
3 - Ready for Review
madsbk
approved these changes
Jun 26, 2024
ayushdg
approved these changes
Jun 26, 2024
|
Yes 👍
…On Wed, 26 Jun 2024 at 21.49, Peter Andreas Entschev < ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In dask_cuda/benchmarks/utils.py
<#1352 (comment)>:
> + if maximum_pool_size is not None:
+ mr = rmm.mr.LimitingResourceAdaptor(
+ mr, allocation_limit=maximum_pool_size
+ )
+
Just to be safe, you mean that we cannot guarantee that maximum_pool_size
is respected because of fragmentation, i.e., if you specify
--maximum-pool-size=4GiB it *may* still go beyond that, is that right?
—
Reply to this email directly, view it on GitHub
<#1352 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAH6FQGDPZI2KOXJXEP557LZJMLMNAVCNFSM6AAAAABJ4TDJKSVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDCNBSHAYDSMBZGI>
.
You are receiving this because your review was requested.Message ID:
***@***.***>
|
raydouglass
approved these changes
Jun 27, 2024
|
Thanks everyone! |
|
/merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Allows disabling RMM in benchmarks via a new option
--disable-rmm. This change makes benchmarks a little more similar to RMM setup inLocalCUDACluster/dask cuda worker, where not specifyingrmm-pool-sizeor specifyingNoneas its value entirely disables setting up RMM as the default allocator. Since for benchmarks it's desired that the default is having an RMM pool we cannot change the default--rmm-pool-sizetoNoneas that would make benchmarks run much slower by default, therefore--disable-rmmis the closest we can make this to the rest of Dask-CUDA.Additionally add
--rmm-maximum-pool-sizefor benchmarks.