OSDI '14 - GPUnet: Networking Abstractions for GPU Programs
GPUnet: Networking Abstractions for GPU Programs Sangman Kim, Seonggu Huh, Yige Hu, Xinya Zhang, and Emmett Witchel, The University of Texas at Austin; Amir Wated and Mark Silberstein, Technion—Israel Institute of Technology Despite the popularity of GPUs in high-performance and scientific computing, and despite increasingly generalpurpose 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. View the full OSDI '14 program at https://www.usenix.org/conference/osdi14/technical-sessions
GPUnet: Networking Abstractions for GPU Programs Sangman Kim, Seonggu Huh, Yige Hu, Xinya Zhang, and Emmett Witchel, The University of Texas at Austin; Amir Wated and Mark Silberstein, Technion—Israel Institute of Technology Despite the popularity of GPUs in high-performance and scientific computing, and despite increasingly generalpurpose 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. View the full OSDI '14 program at https://www.usenix.org/conference/osdi14/technical-sessions