kaushikcfd/feinsum

Database approach for generating near-roofline Einstein Summation kernels.

53
/ 100
Established

Manages code-transformation databases for batched Einstein summations, optimizing memory efficiency by avoiding DRAM flushes of operands. Uses schedule-based transformations recorded from trivial baselines rather than shipping pre-compiled kernels. Integrates with PyTorch and NumPy ecosystems for tensor operations while targeting high-performance compute backends.

Available on PyPI.

Maintenance 13 / 25
Adoption 8 / 25
Maturity 18 / 25
Community 14 / 25

How are scores calculated?

Stars

3

Forks

3

Language

Python

License

MIT

Last pushed

Mar 27, 2026

Monthly downloads

184

Commits (30d)

0

Dependencies

13

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ai-coding/kaushikcfd/feinsum"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.