rtx-50-series

Fast LLM server for NVIDIA Blackwell: GeForce RTX 50 and RTX PRO. Qwen3.8-27B on one RTX 5090: 91-94 tok/s, 1.3x vLLM and 1.4x llama.cpp. Runs the 123.6 GiB Qwen3.8-Flash-Next on a 32 GB card. OpenAI and Anthropic API, tool calls, one docker run.

RTX 5090 & RTX 5060 Docker container with PyTorch + TensorFlow. First fully-tested Blackwell GPU support for ML/AI. CUDA 12.8, Python 3.11, Ubuntu 24.04. Works with RTX 50-series (5090/5080/5070/5060) and RTX 40-series.

Pixal3D ComfyUI integration for Windows (RTX 30/40/50) — single image to textured PBR mesh in 3-5 min

Blackwell-optimized llama.cpp Docker image – works on all NVIDIA GPUs, but tuned for RTX 50 series. Built from scratch with CUDA 12.8, sm_120, NVFP4-ready. 250+ tok/s on 4B F16. Includes llama-chat script.

Debian 13 Trixie - Nvidia Hybrid Graphics Guide

Single-GPU passthrough of an RTX 50 series (Blackwell, 5080 in test case) to a VM on Linux: scripts, EndeavourOS ISO patcher, and the fixes for the black guest screen and the black host display after guest shutdown.

Technical reference data, compatibility notes and reproducible benchmarks for NeuroSK AI Aim Assist.

字幕工坊：Windows 字幕提取、时间轴同步与 SRT/ASS 视频烧录，支持 NVIDIA RTX 50 系列加速。

Space Engineers plugin — fixes the RTX 50-series (Blackwell) grass crash by replacing the geometry-shader foliage pipeline with compute + indirect draw. Same vanilla grass, no geometry shader.

Hunyuan3D-2 fork — image→textured 3D→sliced STL + part segmentation. RTX 50-series (Blackwell/sm_120), CUDA 13.0, Python 3.12, PyTorch 2.11+cu130.

Acha sozinho os melhores pontos de undervolt/overclock/eficiencia da sua GPU NVIDIA (RTX 40xx/50xx)

Sıfırdan yazılmış, bağımlılıksız C++20 + CUDA GPT eğitim motoru. Tek exe, RTX 50-serisi uyumlu.

My personal DLSS 5 gaming configuration guides

GraspM3/DexRep 在 RTX 50 系显卡上的兼容训练、可视化与 Isaac Gym 评估结果