Disag MR3: Add heterogeneous KV/Mamba reshard planners - #5188
Merged
Conversation
Contributor
|
This PR has been automatically converted to draft because all PRs must start as drafts. When you are ready for review, click Ready for Review to begin the review process. This will:
See the contribution guide for more details. |
svcnvidia-nemo-ci
marked this pull request as draft
June 5, 2026 17:36
wdykas
force-pushed
the
dynamo-kv-reshard
branch
9 times, most recently
from
June 8, 2026 17:42
8f7c7c8 to
bfd7580
Compare
santhnm2
approved these changes
Jun 12, 2026
Comment on lines
+44
to
+48
| global_rank: int | ||
| tp_size: int | ||
| tp_rank: int | ||
| layer_start: int # global Mamba-layer index of this rank's first layer | ||
| num_layers: int # Mamba layers held locally (this PP stage) |
Contributor
There was a problem hiding this comment.
Would it make sense to pass a process group object here instead of the individual config fields?
shanmugamr1992
approved these changes
Jun 16, 2026
kvareddy
approved these changes
Jun 16, 2026
wdykas
marked this pull request as ready for review
June 16, 2026 14:38
svcnvidia-nemo-ci
added
Approved
Contributor
Author
|
/ok to test 8651d77 |
Contributor
Author
|
/ok to test cab03e5 |
wdykas
force-pushed
the
dynamo-kv-reshard
branch
from
June 22, 2026 18:49
cab03e5 to
cbb85f8
Compare
Signed-off-by: wdykas <wdykas@nvidia.com>
wdykas
force-pushed
the
dynamo-kv-reshard
branch
from
June 22, 2026 19:02
cbb85f8 to
40b87bd
Compare
Contributor
Author
|
/ok to test 40b87bd |
Contributor
|
🔄 Merge queue validation started! You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/27984695431 |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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.
Pure range-intersection planners that map a request's paged KV (and Mamba conv/ssm state) from a source parallelism to a differently-sharded destination -- the math a disaggregated prefill->decode handoff needs.
What does this PR do ?
Issue tracking
For PRs from open-source community contributors:
Linked issue:
Contribution process
Pre-checks
Code review
Feel free to message or comment @NVIDIA/mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!
All PRs start as draft. If you open a non-draft PR, it will be automatically converted to draft.
Step 1: Mark PR as "Ready for Review"
.github/CODEOWNERS.Final Review might get declined if these requirements are not fulfilled.
Step 2: Final Review
For PRs that change
megatron/core, once all expert reviewers have approved, theFinal Reviewlabel is applied automatically and final reviewers are assigned.For PRs outside
megatron/core, this step is skipped.Step 3: Approved
Once all required reviewers have approved, the
Approvedlabel is applied automatically.Merge
Any member of mcore-engineers will be able to merge your PR.