mirror of https://github.com/harvard-edge/cs249r_book.git synced 2026-07-16 06:07:17 -05:00

Files

Vijay Janapa Reddi ff8c5c0f83 chore(mlsysim): OSS developer QoL improvements and test resilience

- Added 'make docs', 'make docs-preview', and 'make audit' targets to Makefile for easier local development.
- Added comprehensive README.md to the vscode-ext workbench extension.
- Refactored test_engine.py to dynamically import calibration constants rather than hardcoding physics assumptions, ensuring tests don't break if base parameters are tuned.
- Fixed a registry path alias in philosophy.qmd caught by the doc drift linter.

2026-05-25 19:47:11 -04:00

2.1 KiB

Raw Permalink Blame History

MLSysim Workbench (VS Code Extension)

The MLSysim Workbench is the official Visual Studio Code extension for the MLSys·im analytical modeling framework. It provides a deeply integrated, interactive UI for exploring the Silicon Zoo, running evaluations, and searching the design space without leaving your editor.

Features

Zoo Explorer: Browse the built-in Hardware and Models registries directly from the sidebar.
Quick Evaluation: Right-click any hardware or model in the sidebar to run an instant mlsysim eval directly in the integrated terminal.
YAML Scenario Integration: Right-click any .yaml configuration file to evaluate, optimize, or export schemas.
Test Runner: Run the full MLSys·im pytest suite or individual test files with one click.
Action History: Re-run recent evaluations and optimizations from the "Recent Runs" pane.

Requirements

You must have the mlsysim Python package installed in your active environment.

pip install mlsysim

If you are using a virtual environment (venv, conda, etc.), make sure VS Code has the correct Python interpreter selected, or explicitly configure the path in settings.

Extension Settings

This extension contributes the following settings:

mlsysim.pythonPath: Path to the Python interpreter (default: python3).
mlsysim.defaultPrecision: Default precision used for Quick Evals (e.g., fp16).
mlsysim.defaultBatchSize: Default batch size for Quick Evals (default: 1).
mlsysim.defaultEfficiency: Default MFU efficiency for Quick Evals (default: 0.5).
mlsysim.outputFormat: The format for CLI output (text, json, markdown).

Building & Installing Locally

To package and install the extension manually:

Install vsce globally: npm install -g @vscode/vsce
Run npm install inside the vscode-ext directory.
Package the extension: vsce package
Install the generated .vsix file in VS Code:
- View -> Command Palette -> Extensions: Install from VSIX...

Part of the Machine Learning Systems textbook ecosystem.

2.1 KiB Raw Permalink Blame History