dnv2003/v100-skinny

High-performance single-GPU inference for selected model checkpoints and GPUs.

V100 / SM70-focused vLLM engineering fork for modern LLM inference.

NInfer port for Qwen 3.8 27B on RTX 4090

Hyper optimised Qwen3.8-27B inference on one RTX 3090: ReplaySSM, MTP3, reasoning effort, C1-C8 batching, and native Windows and Linux builds.

Qwen3.8-27B on one RTX 4090: full native 262K context via E8 4-bit KV, up to 149 tok/s code decode with MTP3, sm_89-retuned attention prefill, vision, llama.cpp-compatible /metrics + /slots

Deployment kit: Qwen3.8-27B EXL3 3.5bpw target + DFlash2 EXL3 5.0bpw speculative draft — launcher, config, model cards

Qwen3.8-27B on 16-32 GB Nvidia GPUs one-click install for Windows / Linux

~1300 line, minimal, extensible LLM inference engine built from scratch.

Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.