Abstract | ||
---|---|---|
Deep learning has been shown as a successful method for various tasks, and its popularity results in numerous open-source deep learning software tools. Deep learning has been applied to a broad spectrum of scientific domains such as cosmology, particle physics, computer vision, fusion, and astrophysics. Scientists have performed a great deal of work to optimize the computational performance of deep learning frameworks. However, the same cannot be said for I/O performance. As deep learning algorithms rely on big-data volume and variety to effectively train neural networks accurately, I/O is a significant bottleneck on large-scale distributed deep learning training. This study aims to provide a detailed investigation of the I/O behavior of various scientific deep learning workloads running on the Theta supercomputer at Argonne Leadership Computing Facility. In this paper, we present DLIO, a novel representative benchmark suite built based on the I/O profiling of the selected workloads. DLIO can be utilized to accurately emulate the I/O behavior of modern scientific deep learning applications. Using DLIO, application developers and system software solution architects can identify potential I/O bottlenecks in their applications and guide optimizations to boost the I/O performance leading to lower training times by up to 6.7x. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/CCGrid51090.2021.00018 | 2021 IEEE/ACM 21st International Symposium on Cluster, Cloud and Internet Computing (CCGrid) |
Keywords | DocType | ISBN |
deep learning,scientific applications,representative,benchmark,data intensive,I/O,characterization,Tensorflow,data pipeline | Conference | 978-1-7281-9587-2 |
Citations | PageRank | References |
3 | 0.39 | 0 |
Authors | ||
5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Hariharan Devarajan | 1 | 4 | 2.42 |
Huihuo Zheng | 2 | 3 | 0.72 |
Anthony Kougkas | 3 | 4 | 2.76 |
Xian-he Sun | 4 | 1987 | 182.64 |
Venkatram Vishwanath | 5 | 507 | 47.27 |