S
Oct 13, 2026SonicBoom: 52 tok/s decode on a laptop RTX 3050 (llama.cpp: 45) with the same quantized kernels, the difference is a graph compiler. No Python.
I’ve been building SonicBoom, a native tensor runtime written in C++/CUDA with a C ABI and no Python...
Oct 13, 20262 min read0 reactions2 comments