Refactor Mamba chunked prefill

[santhnm2]

What does this PR do ?

Refactors Mamba chunked prefill to treat the first prefill request as the chunked prefill request rather than the last prefill request. Although the context ensures that the last request is always the designated chunked prefill request, what we actually care about is ensuring that any prefill request with non-zero initial states is executed through the chunked prefill path. There are three scenarios:

Scenario A: No prefill request has initial states yet but the last request is the designated chunked prefill request
Scenario B: There is exactly 1 prefill request which is a continuing chunked prefill request with non-zero initial states
Scenario C: There is a leftover chunked prefill request that is executing its last chunk, followed by additional prefill requests

In scenarios B and C the first prefill request has the relevant initial states. The implementation generalizes to scenario A as well where the first prefill request will be treated as if it has non-zero initial states but this is harmless since the initial states are 0.

This refactor allows us to not have to terminate scheduling early to maintain the invariant that the last request is the relevant chunked prefill request (i.e., the request with the non-zero initial states).

⚠️ For major changes (either in lines of code or in its impact), please make sure to first share a design doc with the team. If you're unsure what's the best way to do so, contact the @mcore-oncall.

Contribution process

flowchart LR
    A[Pre-checks] --> B[PR Tests]
    subgraph Code Review/Approval
        C1[Expert Review] --> C2[Final Review]
    end
    B --> C1
    C2 --> D[Merge]

Loading

Pre-checks

• I want this PR in a versioned release and have added the appropriate Milestone (e.g., Core 0.8)
• I have added relevant unit tests
• I have added relevant functional tests
• I have added proper typing to my code Typing guidelines
• I have added relevant documentation
• I have run the autoformatter.sh on my PR

Code review

The following process is enforced via the CODEOWNERS file for changes into megatron/core. For changes outside of megatron/core, it is up to the PR author whether or not to tag the Final Reviewer team.

For MRs into `main` branch

Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!

(Step 1): Add PR label Expert Review

(Step 2): Collect the expert reviewers reviews

1. Attach the Expert Review label when your PR is ready for review.
2. GitHub auto-assigns expert reviewers based on your changes. They will get notified and pick up your PR soon.

⚠️ Only proceed to the next step once all reviewers have approved, merge-conflict are resolved and the CI is passing.
Final Review might get declined if these requirements are not fulfilled.

(Step 3): Final Review

1. Add Final Review label
2. GitHub auto-assigns final reviewers based on your changes. They will get notified and pick up your PR soon.

(Optional Step 4): Cherry-pick into release branch

If this PR also needs to be merged into core_r* release branches, after this PR has been merged, select Cherry-pick to open a new PR into the release branch.

For MRs into `dev` branch

The proposed review process for `dev` branch is under active discussion.

MRs are mergable after one approval by either eharper@nvidia.com or zijiey@nvidia.com.

Merging your PR

Any member of core-adlr and core-nemo will be able to merge your PR.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[kvareddy]

LGTM.

[santhnm2]

/ok to test 2f142a3

[santhnm2]

/ok to test 313d4a3

[santhnm2]

/ok to test f00489e

[santhnm2]

/ok to test 99d98f3

[santhnm2]

/ok to test 758ffe0

[ko3n1g]

🔄 Merge queue validation started!

You can track the progress here: https://github.com/NVIDIA/Megatron-LM/actions/runs/21943803326