随着 AI 使用场景不断扩展（从文档摘要到定制化软件代理），开发者和技术爱好者正在寻求以更 快、更灵活的方式来运行大语言模型（LLM）。 在配备 NVIDIA GeForce RTX GPU 的 PC 上本地运行模型，可实现高性能推理、增强型数据隐私保护，以及对 AI 部署与集成的 ...

NVIDIA近日在CUDA 12.8开发包的更新日志中，针对Maxwell、Pascal及Volta这三款较旧的GPU架构，做出了明确的未来规划。公司宣布，这些架构已完成了全部功能特性的集成，意味着在后续的CUDA版本中，它们将进入“功能冻结”状态。 换句话说，尽管这些老架构仍将继续 ...

When: Monday, Nov. 17, 2008, 8:30 a.m. – 5:00 p.m. Last year's CUDA tutorial at the SC07 conference was a standing-room only event so NVIDIA is pleased to announce that members of its CUDA engineering ...

NVIDIA has announced details of its second annual CUDA tutorial session at the Supercomputing conference. This year, NVIDIA has called in support from the academic and commercial world. John Stone and ...