R

ram-coffers

Project information

LLM infrastructure cost reduction via NUMA-aware weight banking: 147 t/s (8.8x stock llama.cpp) on refurbished enterprise POWER8. Self-hosted inference, no cloud APIs. Part of the Proof of Physical AI stack. (Mirror of github.com/Scottcjn/ram-coffers; full history archived 2026-10-20.)

100 Commits
8 Branches
0 Tags

README
GNU AGPLv3
CHANGELOG
CONTRIBUTING

Created on

October 21, 2026

Loading