Title | ||
---|---|---|
A case study of data transfer efficiency optimization for GPU- and infiniband-based clusters |
Abstract | ||
---|---|---|
Many studies have been performed to realize a high-performance computing environment based on GPU-based clusters. With GPU-based clusters, efficient data transfer is very important because it affects the performance of GPU-based cluster computing systems. GPUDirect RDMA technology is introduced to improve the data transfer efficiency between two nodes. In this paper, we describe a case study of the optimization of the data transfer efficiency of GPUDirect RDMA for GPU- and InfiniBand-based clusters. Based on experimental results, the data transfer latency can be improved by as much as 50% compared to a non-optimized system. |
Year | DOI | Venue |
---|---|---|
2015 | 10.1145/2811411.2811468 | RACS |
Field | DocType | Citations |
Cluster (physics),InfiniBand,Data transmission,Supercomputer,Latency (engineering),Computer science,Parallel computing,Remote direct memory access,Computer cluster | Conference | 1 |
PageRank | References | Authors |
0.48 | 5 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Bongjae Kim | 1 | 4 | 2.25 |
Hyedong Jung | 2 | 6 | 2.00 |