Reverse engineering NVIDIA SASS instruction dictionary, kernel audits and pattern recognition across GPU architectures.

blackwell-architecture 开源项目推荐(8 个)
blackwell-architecture 话题在 GitHub 上没有填写简介,这里按 Star 数汇总最相关的热门仓库。
热门项目 全部 8
GitHub 共 8 个仓库,已全部展示适用于跨平台和CUDA版本的llama-cpp-python预制轮子
A cleaner, modernized reboot of Real-ESRGAN focused on best practices, up-to-date hardware support, and pragmatic developer ergonomics. Supports RTX 50-series (Blackwell) GPUs with automatic device selection, CPU fallback, and improved inference pipelines for both images and videos.
Prebuilt spconv v2.3.8 wheels for CUDA 12.8 / 13.0 with native Blackwell (RTX 50-series, sm_120) kernels — for default PyPI torch (cu130) or torch +cu128
Building PaddlePaddle Inference C++ from source for NVIDIA Blackwell (RTX 50-series, sm_120) with CUDA 13 and TensorRT 10
One-script installer for Microsoft TRELLIS.2 on NVIDIA Blackwell (RTX 5090 / 50-series) under WSL2
A lightweight wrapper for the Chatterbox SoTA TTS providing an OpenAI-compatible API schema.
LLM Inference Optimization for RTX 5070 Ti (Blackwell) - Benchmarks, quantization, serving configs for vLLM/SGLang/TensorRT-LLM/llama.cpp
关于这个话题
- 清单取自 GitHub 话题页并按 Star 数排序,每次展示前 8 个项目;原始页面见 https://github.com/topics/blackwell-architecture。
- 该话题在 GitHub 上共有 8 个仓库, 本站收录的是其中最热门的 8 个,不是全量列表。
- Star / Fork 每日自动刷新一次,实时数据以 GitHub 为准。
- 当前 8 个项目累计 454 Star,覆盖 3 种主要语言,数据更新于 2026-11-11。
- 中文介绍来自 GitHub 官方话题说明(github/explore)的机器翻译,可点「查看原文」对照;译文缺失时会先展示原文。