ProprietaryLegal/Volta-sPLIt-DeepSeek-V4.1-Flash-1017-8xV100-SXM2-PP2xTP4
DeepSeek-V4.1-Flash-1017 on 8x V100-SXM2 (2018): 119.0 tok/s decode, 120.8 thinking on (measured), 524K context, 300K-token first token in 178 s. PP2xTP4 port of 1Cat-vLLM #1277 with every patch, measurement and lost lever. PLI Labs.
GitHub repository with 30 stars and 3 forks.
Language: Python
Topics: cuda, deepseek, gguf, llm-inference, speculative-decoding, v100, vllm, volta