Extending Tensor Parallelism for IBM FMS: Sequence Parallelism#455
Open
Sibi-Git wants to merge 48 commits into
Open
Extending Tensor Parallelism for IBM FMS: Sequence Parallelism#455Sibi-Git wants to merge 48 commits into
Sibi-Git wants to merge 48 commits into
Conversation
…h DTensor API for Llama and Granite
Automate tensor parallel plan generation, sequence parallelismn supported
- Added latency and memory usage plots for L40 GPU and Xeon CPU tests. - Included execution tables and brief analysis under Results and Benchmarks section in README. - Added Wandb link
added cpu test scripts
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This project extends the IBM Foundation Model Stack (FMS) to support both Tensor Parallelism (TP) and Sequence Parallelism (SP) in distributed model inference. While TP enables parameter sharding across GPUs, it does not partition the sequence dimension, which leads to memory inefficiency at long sequence lengths. We address this by integrating SP into normalization layers, optimizing layout transitions, and enabling support for non-divisible and short sequence lengths.
Members: