Torch integration #692
Open
Torch integration #692
Conversation
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
Contributor
There was a problem hiding this comment.
Pull request overview
This PR adds PyTorch integration to MSCCL++ through a comprehensive refactoring that introduces a new algorithm abstraction layer. The changes enable users to define custom collectives using both DSL (domain-specific language) and native CUDA/HIP kernels, with flexible algorithm selection at runtime.
Key changes include:
- New algorithm abstraction with
Algorithm,NativeAlgorithm, andDslAlgorithmclasses replacing the oldExecutionPlanRegistry - Python bindings for algorithm execution and compilation
- Multiple built-in allreduce and allgather algorithm implementations
- Enhanced error handling with better diagnostic messages
- Support for FP8 data types and various reduce operations
Reviewed changes
Copilot reviewed 58 out of 61 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| include/mscclpp/algorithm.hpp | Major refactor: new Algorithm interface with NativeAlgorithm and DslAlgorithm implementations |
| src/algorithms/algorithm.cc | Implementation of algorithm framework with default algorithm builders |
| src/include/algorithms/utils.hpp | New utility header for algorithm setup (missing license) |
| src/include/algorithms/allreduce/common.hpp | Comprehensive allreduce operations with FP8 support |
| src/algorithms/allreduce/*.cu | Multiple allreduce algorithm implementations |
| src/algorithms/allgather/*.cu | Allgather algorithm implementations |
| python/mscclpp/_algorithm.py | Python wrapper for algorithm execution |
| python/mscclpp/_compiler.py | DSL and native code compilation support |
| python/csrc/algorithm.cpp | Python C++ bindings for algorithms |
| examples/torch-integration/*.py | Example integration with PyTorch |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
examples/torch-integration/customized_comm_with_default_algo.py
Outdated
Show resolved
Hide resolved
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Contributor
Author
|
/azp run |
|
Azure Pipelines successfully started running 3 pipeline(s). |
Binyang2014
requested review from
caiomcbr,
chhwang,
mahdiehghazim and
seagater
Contributor
Author
|
/azp run |
|
Azure Pipelines successfully started running 3 pipeline(s). |
Contributor
Author
|
/azp run |
|
Azure Pipelines successfully started running 3 pipeline(s). |
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.
Reorganize current native algorithm implementation and DSL algorithm implementation.
Provide unified API for DSL algo and native algo and provide interface to tune the algo
Provide interface for pytorch integration with native API and DSL