Sudhu/megatron ci benchmarking#111
Draft
sudhu2k wants to merge 14 commits into
Draft
Conversation
added 3 commits
February 12, 2026 18:52
- Introduced `run_benchmarks.sh` to execute Megatron-LM benchmarks and generate performance reports in both text and JSON formats. - Updated GitHub Actions workflow to run benchmarks, display results, and upload benchmark artifacts. - Ensured compatibility with existing CI setup by handling errors gracefully and logging outputs for review.
…orting - Modified `run_benchmarks.sh` to support 70B model configurations for various training setups. - Introduced a new Python script `benchmark-report-to-markdown.py` to convert benchmark results into a markdown table format. - Enhanced GitHub Actions workflow to include a dedicated job for running unit tests and generating benchmark reports, with improved artifact handling and reporting. - Cleaned up existing scripts by removing unnecessary comments and ensuring consistent formatting.
|
This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days. |
|
This PR was closed because it has been inactive for 7 days since being marked as stale. |
- Updated `run_benchmarks.sh` to allow selective benchmarking via `BENCHMARK_SUITE` input, enabling targeted runs for Llama and DeepSeek models. - Introduced `run_profile_ci.sh` for profiling training runs with PyTorch, including optional TraceLens report generation. - Added new tools for generating memory footprint and performance breakdown charts from benchmark logs. - Enhanced CI workflows to support optional benchmark chart generation and profiling steps, improving performance reporting and analysis. - Refactored existing scripts for clarity and consistency, ensuring better integration with CI processes.
Made-with: Cursor # Conflicts: # .github/workflows/megatron-ci.yml
- Added functionality to write the Docker image to a tar file, allowing downstream jobs to load the image without pulling from Docker Hub. - Implemented artifact upload and download steps for the Docker image, improving reliability on runners with connectivity issues. - Cleaned up temporary image files after tests to maintain a tidy workspace.
…in step and utilize a read-only registry cache for building images. This change enhances the workflow by ensuring proper authentication and improving caching efficiency during benchmarking.
- Updated Dockerfile to include installation of openpyxl for Excel report generation. - Modified `run_benchmarks.sh` to generate TraceLens Excel and text summaries from PyTorch profiler JSON. - Enhanced `run_profile_ci.sh` to run TraceLens report generation and NCCL analysis, outputting results to specified directories. - Added new tools for generating performance breakdown markdown and handling multi-rank NCCL trace analysis. - Improved existing scripts for better clarity and functionality in profiling and reporting processes.
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
1 participant
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.
Motivation
Technical Details
Test Plan
Test Result
Submission Checklist