What happens when you run a CUDA kernel?(fergusfinn.com)273 points by mezark 3 months ago | 31 comments
Speculative KV coding: losslessly compressing KV cache by up to ~4×(fergusfinn.com)137 points by kkm 4 months ago | 28 comments
Bringing Up DeepSeek-V4-Flash on AMD MI300X(fergusfinn.com)107 points by kkm 4 months ago | 17 comments
Also-RANS: Asymmetric Numeral Systems for Entropy Coding(fergusfinn.com)12 points by mezark 5 months ago | 0 comments