-
Notifications
You must be signed in to change notification settings - Fork 257
Profiling ExecuTorch models with SME2 acceleration #2791
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
pareenaverma
merged 17 commits into
ArmDeveloperEcosystem:main
from
jzhzhu:sme2-executorch-clean
Jan 23, 2026
Merged
Profiling ExecuTorch models with SME2 acceleration #2791
pareenaverma
merged 17 commits into
ArmDeveloperEcosystem:main
from
jzhzhu:sme2-executorch-clean
Jan 23, 2026
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
- Add complete profiling pipeline for ExecuTorch models with SME2 acceleration - Include model-agnostic profiling scripts and configs - Add agent skills for automation - Include example models and test cases - Add comprehensive documentation with operator-level profiling insights
- Complete learning path with 4 main pages (overview, setup, model onboarding, agent skills) - Model-agnostic profiling pipeline with config-driven experiments - Agent skills for AI assistant automation (7 skills total) - Updated scripts: analyze_results.py (CSV generation, optional etrecord) - New scripts: generate_report.py (markdown reports from CSV) - New tools: etdump_to_csv.py, generate_kernel_view.py - Consistent output structure: out_<model>/artifacts/ and out_<model>/runs/ - Removed unused script: compare_run_to_known_good.py - Updated all documentation to reflect new structure and workflow - Fixed typos, path inconsistencies, and script parameter updates - Added PIPELINE_COMMANDS.md reference documentation
… requirements - Rename SME2_stack_01062026.png -> sme2_stack_01062026.png - Rename PIPELINE_COMMANDS.md -> pipeline_commands.md - Rename CMakePresets.json -> cmake_presets.json (asset file only) - Update all references in markdown files and Python scripts - ExecuTorch's CMakePresets.json remains unchanged (standard CMake filename)
…ents - Rename executorch_sme2_kit/README.md -> readme.md - Rename executorch_sme2_kit/agent_skill_ml_profiling/README.md -> readme.md - Rename executorch_sme2_kit/model_profiling/scripts/README.md -> readme.md - Update all references in markdown files
- Updated 08_onboard_edgetam.md to clarify EdgeTAM has multiple modules (only image encoder is onboarded) - Fixed test case file (01-smoke-test-mac.md) to use correct paths (out_<model>/runs/, executorch/cmake-out/) - Removed deprecated --model parameter from analyze_results.py examples in test case - Verified all web content pages, agent skills, and README files are consistent - All file paths use executorch/cmake-out/ and out_<model>/runs/ structure - No linter errors - All content ready for use
… and add to .gitignore
…minology - Update title to 'Revealing latent ExecuTorch latency after SME2 acceleration' - Replace 'profiling' with 'performance analysis' throughout narrative content - Add comprehensive Android device power management settings: - Unconstrained boost mode (value 15) for stress testing hardware - App developer mode (value 1) for real user experience - UI animation settings for consistent measurements - Fix prerequisites YAML formatting (quote Optional field) - Clarify workflow: setup_workspace required before validate_workflow - Update all references to use accurate terminology and settings
- Restore 48 image files that were inadvertently deleted in commit 0816347 - These files belong to other learning paths and should match upstream/main - Also includes latest updates to sme-executorch-profiling content
cd13a23 to
1035d23
Compare
- Remove executorch_sme2_kit/ folder (moved to sme-executorch-profiling repo) - Update all GitHub links to point to new repository: - 01-overview.md: Updated clone instructions and script links - 02-setup-and-pipeline.md: Updated script and asset links - 03-model-onboarding-and-profiling.md: Updated model, script, and agent skill links - 04-agent-skills.md: Updated agent skill links - Add full GitHub URLs to agent skill references for better navigation - Clarify path explanations for consistency - Code is now hosted at: https://github.com/ArmDeveloperEcosystem/sme-executorch-profiling
- Add full GitHub URLs to 08_onboard_edgetam.md references for better navigation - Clarify path explanation for consistency
Contributor
|
merging into main for tech review |
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.
Before submitting a pull request for a new Learning Path, please review Create a Learning Path
Please do not include any confidential information in your contribution. This includes confidential microarchitecture details and unannounced product information.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of the Creative Commons Attribution 4.0 International License.