Changes to support the v8 interface of the NCCL plugin interface #365
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
Sync with headers from upstream NCCL 2.20.5 and pull in the v8 definitions. Signed-off-by: Raghu Raja <raghunch@amazon.com>
Commit 10383ce ("Split api and core source files") introduced nccl_net_ofi_regMr_sizet() with exact same code as nccl_net_ofi_regMr() besides receiving the size argument as size_t rather than int. NCCL net_plugin v8 changes the regMr() API to have a size_t size argument simialr to nccl_net_ofi_regMr_sizet(). Change standard nccl_net_ofi_regMr() to have a size_t size argument and introduce a v4-v7 compatible API that have a int size argument but just calls the standard API after casting the size argument. Remove nccl_net_ofi_regMr_sizet() from the code as it is same as v4-v7 compatible regMr() API. Signed-off-by: Liran Alon <liran@amazon.com>
rajachan
force-pushed
the
move-up-to-v8
branch
2 times, most recently
from
6bc417d to
8b1777b
Compare
rauteric
requested changes
Mar 21, 2024
Two key changes: - regMr size changed from int -> size_t - A new `regIsGlobal` property which is used by NCCL to determine support for user registrations. The plugin now determines this via the mr_mode bit providers set to define the scope of a MR (domain-level or endpoint-level). Signed-off-by: Raghu Raja <raghunch@amazon.com> Signed-off-by: Liran Alon <liran@amazon.com> Signed-off-by: Raghu Raja <raghunch@amazon.com>
rauteric
approved these changes
Mar 21, 2024
rajachan
added
the
BuildTriggerRequest
bwbarrett
approved these changes
Mar 26, 2024
rauteric
added
BuildTriggerRequest
bwbarrett
added
BuildTriggerRequest
rajachan
added
BuildTriggerRequest
rauteric
added
BuildTriggerRequest
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.
A more succint version of #345
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.