weicj/vLLM-2080Ti-Definitive

llama.cpp fork for significantly improved performance on Ampere (especially RTX 3090 / 3090 Ti): example: 95+ tok/s over a 100K-token generation at temperature 1 for qwen3.8, supporting full 262K ctx. Supports low memory (12gb) configs for 200k+ ctx

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

Tile-Based Runtime for Ultra-Low-Latency LLM Inference

Community recipes for serving LLMs on RTX 3090/4090/5090 CUDA gpus. Multi-engine (vLLM, sglang, llama.cpp and other custom engines) and model-agnostic. Currently shipping Qwen3.8-27B Qwen3.6 35B Gemma 4 26B Gemma 4 31B configs for 1×/2×/4x/8x cards.

RTX 6000 Pro Wiki — Running Large LLMs (Qwen3.5-397B, Kimi-K2.5, GLM-5) on PCIe GPUs without NVLink

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

A fast inference library for running LLMs locally on modern consumer-class GPUs

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

NInfer port for Qwen 3.8 27B on RTX 4090