As the computational trend diverges from standard CPU computing, to encompass GPUs and other accelerators, the need to integrate these unused resources within existing systems becomes apparent. This paper presents the implementation of an HTCondor pool with GPU execution capabilities through OpenCL. Implementation is discussed from both the system setup and the software design standpoint. The GPU landscape is investigated and the efficiency of the system is evaluated via FastFourier Transform computations. Experimental results show that HTCondor GPU performance matches a dedicated GPU cluster.
|Title of host publication||Proceedings of the EMerging Technology (EMiT) Conference 2016|
|Editors||B.D. Rogers, D. Topping, F. Mantovani, M.K. Bane|
|Place of Publication||Manchester|
|Publisher||The Emerging Technology (EMiT) Conference|
|Number of pages||4|
|Publication status||Published - 3 Jun 2016|
|Event||Emerging Technology Conference: Emerging Technology: Revolutionising Computing! - Barcelona Supercomputing Center, Barcelona, Spain|
Duration: 2 Jun 2016 → 3 Jun 2016
https://emit.tech/emit-2016/emit2016-intro/ (Link to Conference Website )
|Conference||Emerging Technology Conference|
|Abbreviated title||EMiT 2016|
|Period||2/06/16 → 3/06/16|
Dafinoiu, A., Higgins, J., & Holmes, V. (2016). Accelerating High-Throughput Computing through OpenCL. In B. D. Rogers, D. Topping, F. Mantovani, & M. K. Bane (Eds.), Proceedings of the EMerging Technology (EMiT) Conference 2016 (pp. 46-49). Manchester: The Emerging Technology (EMiT) Conference.