DeepSeek Open Source DeepGEMM – FP8 GEMM Library(300 lines for 1350+ FP8 TFLOPS)
1–2 of 2 posts
Re: DeepSeek Open Source DeepGEMM – FP8 GEMM Library(300 lines for 1350+ FP8 TFLOPS)
#2Github: https://github.com/deepseek-ai/DeepGEMM
- Up to 1350+ FP8 TFLOPS on Hopper GPUs - No heavy dependency, as clean as a tutorial - Fully Just-In-Time compiled - Core logic at ~300 lines - yet outperforms expert-tuned kernels across most matrix sizes - Supports dense layout and two MoE layouts