Skip to content
#

performance-profiling

Here are 14 public repositories matching this topic...

This repository documents a DOI-backed profiling and optimization study of AniSOAP's NumPy and PyTorch descriptor pipelines across CPU and Apple MPS. It was developed during Fall 2025 OSPO internship with the Cersonsky Lab at UW–Madison. It's maintained successor is AniSOAP-Torch.

  • Updated Jul 18, 2026
  • Python

High-precision Python tool to monitor and log CPU, GPU, and RAM energy consumption in real time. Analyze component-level power usage to optimize performance and efficiency. Lightweight, configurable, and ideal for energy-aware development and profiling.

  • Updated Jan 12, 2026
  • Python

Add this topic to your repo

To associate your repository with the performance-profiling topic, visit your repo's landing page and select "manage topics."

Learn more