TY - GEN
T1 - GPU Adaptive In-situ Parallel Analytics (GAP)
AU - Xing, Haoyuan
AU - Agrawal, Gagan
AU - Ramnath, Rajiv
N1 - Publisher Copyright:
© 2022 Association for Computing Machinery.
PY - 2022/10/8
Y1 - 2022/10/8
N2 - Despite the popularity of in-situ analytics in scientifc computing, there is only limited work to date on in-situ analytics for simulations running on GPUs. Notably, two unaddressed challenges are 1) performing memory-efcient in-situ analysis on accelerators and 2)automatically choosing the processing resources and suitable data representation for a given query and platform. This paper addresses both problems. First, GAP makes several new contributions toward making bitmap indices suitable, effective, and efcient as a compressed data summary structure for the GPUs this includes introducing a layout structure, a method for generating multi-attribute bitmaps, and novel techniques for bitmap-based processing of major operators that comprise complex data analytics. Second, this paper presents a performance modeling methodology, aiming to predict the placement (i.e., CPU or GPU) and the data representation choice (summarization or original) that yield the best performance on a given confguration. Our extensive evaluation of complex in-situ queries and real-world simulations shows that with our methods, analytics on GPU using bitmaps almost always outperforms other options, and the GAP performance model predicts the optimal placement and data representation for most scenarios.
AB - Despite the popularity of in-situ analytics in scientifc computing, there is only limited work to date on in-situ analytics for simulations running on GPUs. Notably, two unaddressed challenges are 1) performing memory-efcient in-situ analysis on accelerators and 2)automatically choosing the processing resources and suitable data representation for a given query and platform. This paper addresses both problems. First, GAP makes several new contributions toward making bitmap indices suitable, effective, and efcient as a compressed data summary structure for the GPUs this includes introducing a layout structure, a method for generating multi-attribute bitmaps, and novel techniques for bitmap-based processing of major operators that comprise complex data analytics. Second, this paper presents a performance modeling methodology, aiming to predict the placement (i.e., CPU or GPU) and the data representation choice (summarization or original) that yield the best performance on a given confguration. Our extensive evaluation of complex in-situ queries and real-world simulations shows that with our methods, analytics on GPU using bitmaps almost always outperforms other options, and the GAP performance model predicts the optimal placement and data representation for most scenarios.
UR - http://www.scopus.com/inward/record.url?scp=85147332000&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85147332000&partnerID=8YFLogxK
U2 - 10.1145/3559009.3569661
DO - 10.1145/3559009.3569661
M3 - Conference contribution
AN - SCOPUS:85147332000
T3 - Parallel Architectures and Compilation Techniques - Conference Proceedings, PACT
SP - 467
EP - 480
BT - PACT 2022 - Proceedings of the 2022 International Conference on Parallel Architectures and Compilation Techniques
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 31st International Conference on Parallel Architectures and Compilation Techniques, PACT 2022
Y2 - 8 October 2022 through 10 October 2022
ER -