DeepSeek DeepGEMM: High-Performance GPU BLAS and MoE Kernel Library for SM90 and SM100
DeepSeek open-sources DeepGEMM, a clean GPU tensor core kernel library featuring runtime JIT compilation without pre-builds, FP8/FP4/BF16 GEMMs, communication-o
#DeepGEMM
TAU HOME stories tagged DeepGEMM.