Welcome to AI Power Meter’s documentation!

This library enables to easily monitor energy usage of machine learning programs. It uses RAPL for the CPU and nvidia-smi for the GPU.

• Quick start
  • Hardware requirements
  • Installation
  • Measuring my first program
• Background on power measure
  • General considerations
  • Preliminaries
  • CPU and RAPL
  • GPU and nvidia-smi
  • Measuring multiple programs
  • Related work on power meter libraries
• Advanced use
  • Recorded fields
  • Monitoring whole machine with Prometheus
  • model complexity
  • Docker integration
• Dev Documentation
  • Experiment module
  • rapl_power module
  • gpu_power module
• Deep learning benchmark
  • Summary : One inference with classic deep learning models
  • Experimental protocol
  • Alexnet study
  • Resnet study
  • Bert Transformer
  • Deep rewiring
  • Pruning
  • SNIP
  • Force
• Machine and Deep Learning Benchmarks with wattmeters
  • OmegaWatt Power meters
  • Tracking at Low Frequency from wifi
  • Track with High Frequency measures
  • Benchmarks
• Bibliography

Indices and tables

• Index

• Module Index

• Search Page