Volume 1, Issue 4, November 2013, Pages 196 - 212
A Checkpoint/Restart Scheme for CUDA Programs with Complex Computation States
Hai Jiang, Yulu Zhang, Jeff Jennes, Kuan-Ching Li
Received 20 February 2013, Accepted 15 July 2013, Available Online 1 November 2013.
- https://doi.org/10.2991/ijndc.2013.1.4.2How to use a DOI?
- GPU, CUDA, checkpoint/start, fault tolerance
- Checkpoint/restart has been an effective mechanism to achieve fault tolerance for many long-running scientific applications. The common approach is to save computation states in memory and secondary storage for execution resumption. However, as the GPU plays a much bigger role in high performance computing, there is no effective checkpoint/restart scheme yet due to the difficulty of the GPU computation state handling. This paper proposes an application-level checkpoint/restart scheme to save and restore GPU computation states in annotated user programs. A pre-compiler and run-time support module are developed to construct and save states in CPU system memory dynamically, whereas secondary storage can be utilized for scalability and long-term fault tolerance. CUDA programs with complicated computation states are supported. State-related variables dissipated in various memory units are collected. Both stack and heap are duplicated at application level for state construction. Experimental results have demonstrated the effectiveness of the proposed scheme.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - JOUR AU - Hai Jiang AU - Yulu Zhang AU - Jeff Jennes AU - Kuan-Ching Li PY - 2013 DA - 2013/11 TI - A Checkpoint/Restart Scheme for CUDA Programs with Complex Computation States JO - International Journal of Networked and Distributed Computing SP - 196 EP - 212 VL - 1 IS - 4 SN - 2211-7946 UR - https://doi.org/10.2991/ijndc.2013.1.4.2 DO - https://doi.org/10.2991/ijndc.2013.1.4.2 ID - Jiang2013 ER -