Haupt-Navigation

Page Content

Publications

Pipelined Query Processing in Coprocessor Environments
Citation key	FunkeBNMT18
Author	Henning Funke Sebastian Breß Stefan Noll Volker Markl Jens Teubner
Pages	1603-1618
Year	2018
DOI	https://dl.acm.org/citation.cfm?doid=3183713.3183734
Journal	In International Conference on Management of Data (SIGMOD 2018)
Abstract	Query processing on GPU-style coprocessors is severely limited by the movement of data. With teraflops of compute throughput in one device, even high-bandwidth memory cannot provision enough data for a reasonable utilization. Query compilation is a proven technique to improve memory efficiency. However, its inherent tuple-at-a-time processing style does not suit the massively parallel execution model of GPU-style coprocessors. This compromises the improvements in efficiency offered by query compilation. In this paper, we show how query compilation and GPU-style parallelism can be made to play in unison nevertheless. We describe a compiler strategy that merges multiple operations into a single GPU kernel, thereby significantly reducing bandwidth demand. Compared to operator-at-a-time, we show reductions of memory access volumes by factors of up to 7.5x resulting in shorter kernel execution times by factors of up to 9.5x.

Link to publication [1] Link to original publication [2] Download Bibtex entry [3]

Back [4]

[5]

------ Links: ------

[1] http://www.redaktion.tu-berlin.de/fileadmin/fg131/P
ublikation/Papers/Funke_pipelined-query-processing_2018
_preprint.pdf

[2] https://dl.acm.org/citation.cfm?doid=3183713.318373
4

[3] https://www.dima.tu-berlin.de/menue/publications/pu
blications/parameter/en/font4/maxhilfe/?no_cache=1&
tx_sibibtex_pi1%5Bdownload_bibtex_uid%5D=1422296&tx
_sibibtex_pi1%5Bcontentelement%5D=tt_content%3A126920

[4] https://www.dima.tu-berlin.de/menue/publications/pu
blications/parameter/en/font4/maxhilfe/

[5] https://www.dima.tu-berlin.de/menue/publications/bi
g_data_management_report/parameter/en/font4/maxhilfe/