Skip to content

feat: add EcoTrace - hardware-level carbon and energy measurement for Python - #3400

Open
Zwony wants to merge 1 commit into
vinta:masterfrom
Zwony:master
Open

Zwony wants to merge 1 commit into
vinta:masterfrom
Zwony:master

Conversation

@Zwony

@Zwony Zwony commented Nov 5, 2026

Copy link
Copy Markdown

Description

Adds EcoTrace to the Science > Sustainability subsection.

EcoTrace is a high-precision Python library for measuring exact energy consumption and CO? emissions of Python applications and AI/LLM workloads. Unlike estimation-based tools, it reads directly from hardware counters (Intel RAPL, NVML, Apple Silicon) with process-scoped isolation.

Why it belongs here:

  • Pure Python, \pip install ecotrace, zero configuration
  • MIT licensed, 100% offline and local-first
  • Hardware-level telemetry: RAPL / NVML / Apple Silicon
  • Official LangChain integration (catalogued in LangChain Integration Directory)
  • Native HuggingFace Trainer callback
  • Zenodo DOI: 10.5281/zenodo.22834060
  • OpenSSF Best Practices certified

Signed-off-by: Emre Ozkal ecotraceteam@gmail.com

Signed-off-by: Emre Ozkal <ecotraceteam@gmail.com>
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
Comment thread README.md Resolved
@JinyangWang27

Copy link
Copy Markdown
Collaborator

@Zwony Thanks for the submission. I'm considering adding a Sustainability subsection. Could you point to evidence of academic use of EcoTrace, such as papers that cite it or research groups that use it? Is there a publication about the project itself? Zenodo does not count

@Zwony

Zwony commented Nov 16, 2026

Copy link
Copy Markdown
Author

Hi @JinyangWang27,

Thank you! Having a dedicated Sustainability subsection would be a fantastic addition to awesome-python as green computing and software carbon accounting gain global traction.

To be completely transparent: Our current publication is our own Technical Whitepaper and benchmark evaluation (docs/WHITE_PAPER.md). We have recently initiated formal outreach and collaboration with university High-Performance Computing & Systems research groups to conduct physical wattmeter validation and co-author a peer-reviewed academic paper.

Because the formal peer-review and publication cycle will naturally take a few months, here is my suggestion:

  1. For now, we can list EcoTrace under "Debugging Tools -> Profiler" (alongside memray, py-spy, and scalene), where it fits naturally today as a process-level energy and hardware profiler.
  2. Once our peer-reviewed research paper is officially published with formal citations, I will open a follow-up PR with the academic DOI/paper to inaugurate the Sustainability section.

If this staged approach works for you, I'll gladly update the PR diff to place it under Debugging Tools -> Profiler!

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants