NVIDIA / NVIDIA/Megatron-LM

MFSDP v2: overlap DP-inner and DP-outer communication

Open
#6,714 15 comments 0 reactions 0 assignees View on GitHub
MFSDPv2 module: megatron-fsdp
Dominant language
Python
Stars
17.9k
Forks
4.5k
Avg merge
4d 6h
Merged PRs (30d)
271

Description

Follow-up to #5616: explore overlapping DP-inner and DP-outer communication in MFSDP v2.

The two communications use different network domains, and early experience suggests that overlapping them could reduce latency significantly. The implementation should follow the runtime scheduling model in `megatron/core/distributed/fsdp/src/docs/runtime_schedule.md`.

We should first measure the benefit and use those results to guide the implementation and scheduling design.

Contributor guide

Open the contributing guide

Research direction

Read megatron/core/distributed/fsdp/src/docs/runtime_schedule.md and review the MFSDP v2 communication paths described in follow-up issue #5616. First measure whether DP-inner and DP-outer communication overlap reduces latency, then use those results to define the scheduling design and implementation scope.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
distributed-systems, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.