https://github.com/dannixon/yuugi

https://github.com/dannixon/yuugi

energy-consumption hacktoberfest power-consumption prometheus

Last synced: 7 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/dannixon/yuugi
• Owner: DanNixon
• License: mit
• Created: 2022-08-03T10:49:02.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2024-03-05T09:25:17.000Z (8 months ago)
• Last Synced: 2024-03-06T00:25:53.695Z (8 months ago)
• Topics: energy-consumption, hacktoberfest, power-consumption, prometheus
• Language: Rust
• Homepage:
• Size: 60.5 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
# yuugi [![CI](https://github.com/DanNixon/yuugi/actions/workflows/ci.yml/badge.svg)](https://github.com/DanNixon/yuugi/actions/workflows/ci.yml) [![dependency status](https://deps.rs/repo/github/dannixon/yuugi/status.svg)](https://deps.rs/repo/github/dannixon/yuugi)

A quick, hacky proof of concept tool for monitoring power consumption of multi-component software systems.

See `yuugi --help` for details of configurable options.

Currently only provides a vague estimate of power consumption due to horrific assumptions about the state of the system's CPU but is better than nothing and using process CPU time as a proxy measurement is still valid to identifiy possible optimisation targets.

The following caveats apply to using this:

- CPU power must be manually derived (if in doubt the TDP will be a sensible guess, this will in most cases result in an overestimation of power consumption)

- CPU frequency/power states are not taken into account (if a core is halted or running at a lower frequency then power consumption will be overestimated)

- Mutliple socket systems will probably not "just work"

- Short lived processes will not be reported correctly if their life is less than or not significantly longer than the polling interval

- The current label set for CPU time and energy generates a *lot* of series (be sure whatever collection system to feed this into is suitable to do so)

TL;DR: probably just use the CPU time measurement.