Journal
ACM SIGPLAN NOTICES
Volume 52, Issue 8, Pages 235-248Publisher
ASSOC COMPUTING MACHINERY
DOI: 10.1145/3018743.3018756
Keywords
Multi-GPU; Asynchronous Programming; Irregular Algorithms
Categories
Funding
- German Research Foundation (DFG) Priority Program Software for exascale Computing (SPP-EXA), research project FFMK [1648]
- NSF [1218568, 1337281, 1406355, 1618425]
- DARPA BRASS [750-16-2-0004]
- NVIDIA
- Direct For Computer & Info Scie & Enginr
- Division of Computing and Communication Foundations [1218568] Funding Source: National Science Foundation
- Division Of Computer and Network Systems
- Direct For Computer & Info Scie & Enginr [1406355] Funding Source: National Science Foundation
- Division of Computing and Communication Foundations
- Direct For Computer & Info Scie & Enginr [1337281] Funding Source: National Science Foundation
Ask authors/readers for more resources
Nodes with multiple GPUs are becoming the platform of choice for high-performance computing. However, most applications are written using bulk-synchronous programming models, which may not be optimal for irregular algorithms that benefit from low-latency, asynchronous communication. This paper proposes constructs for asynchronous multi-GPU programming, and describes their implementation in a thin runtime environment called Groute. Groute also implements common collective operations and distributed work-lists, enabling the development of irregular applications without substantial programming effort. We demonstrate that this approach achieves state-of-the-art performance and exhibits strong scaling for a suite of irregular applications on 8-GPU and heterogeneous systems, yielding over 7x speedup for some algorithms.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available