Despite the popularity of GPUs in high-performance and scientific computing, and despite increasingly general-purpose hardware capabilities, the use of GPUs in network servers or distributed systems poses significant challenges. GPUnet is a native GPU networking layer that provides a socket abstraction and high-level networking APIs for GPU programs. We use GPUnet to streamline the development of high-performance, distributed applications like in-GPU-memory MapReduce and a new class of low-latency, high-throughput GPU-native network services such as a face verification server.

中文翻译：

---

GPU网

尽管 GPU 在高性能和科学计算中很受欢迎，并且尽管硬件功能越来越通用，但在网络服务器或分布式系统中使用 GPU 带来了重大挑战。GPUnet 是一个原生 GPU 网络层，它为 GPU 程序提供套接字抽象和高级网络 API。我们使用 GPUnet 来简化高性能分布式应用程序的开发，例如 GPU 内存中的 MapReduce 和一类新的低延迟、高吞吐量 GPU 原生网络服务，例如人脸验证服务器。