deepseek-ai/DeepGEMM

DeepGEMM: clean and efficient BLAS kernel library on GPU

391 best points · 67 comments · 1 Hacker News post · 9,159 GitHub stars · View on GitHub

Hacker News discussions

• DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling (391 points, 67 comments, by mfiguiere, 2025-02-26)