Scipediacontent: Scipediacontent moved page Draft Content 333643256 to Passing et al 2019a

2021-01-28T16:48:22Z

Scipediacontent moved page Draft Content 333643256 to Passing et al 2019a

Scipediacontent: Created page with " == Abstract == Increasing single instruction multiple data (SIMD) capabilities in modern hardware allows for the compilation of data-parallel query pipelines. This means GPU..."

2021-01-28T16:48:20Z

Created page with " == Abstract == Increasing single instruction multiple data (SIMD) capabilities in modern hardware allows for the compilation of data-parallel query pipelines. This means GPU..."

New page

== Abstract ==

Increasing single instruction multiple data (SIMD) capabilities in modern hardware allows for the compilation of data-parallel query pipelines. This means GPU-alike challenges arise: control flow divergence causes the underutilization of vector-processing units. In this paper, we present efficient algorithms for the AVX-512 architecture to address this issue. These algorithms allow for the fine-grained assignment of new tuples to idle SIMD lanes. Furthermore, we present strategies for their integration with compiled query pipelines so that tuples are never evicted from registers. We evaluate our approach with three query types: (i) a table scan query based on TPC-H Query 1, that performs up to 34% faster when addressing underutilization, (ii) a hashjoin query, where we observe up to 25% higher performance, and (iii) an approximate geospatial join query, which shows performance improvements of up to 30%.

Document type: Conference object

== Full document ==
<pdf>Media:Draft_Content_333643256-beopen231-7032-document.pdf</pdf>

== Original document ==

The different versions of the original document can be found in:

* [https://ir.cwi.nl/pub/27874 https://ir.cwi.nl/pub/27874]

* [https://ir.cwi.nl/pub/27874/27874.pdf https://ir.cwi.nl/pub/27874/27874.pdf]

* [https://ir.cwi.nl/pub/28788 https://ir.cwi.nl/pub/28788]

* [https://link.springer.com/content/pdf/10.1007/s00778-019-00547-y.pdf https://link.springer.com/content/pdf/10.1007/s00778-019-00547-y.pdf] under the license cc-by

* [http://dl.acm.org/ft_gateway.cfm?id=3211928&ftid=1978228&dwn=1 http://dl.acm.org/ft_gateway.cfm?id=3211928&ftid=1978228&dwn=1],
: [http://dx.doi.org/10.1145/3211922.3211928 http://dx.doi.org/10.1145/3211922.3211928] under the license http://www.acm.org/publications/policies/copyright_policy#Background

* [http://link.springer.com/content/pdf/10.1007/s00778-019-00547-y.pdf http://link.springer.com/content/pdf/10.1007/s00778-019-00547-y.pdf],
: [http://link.springer.com/article/10.1007/s00778-019-00547-y/fulltext.html http://link.springer.com/article/10.1007/s00778-019-00547-y/fulltext.html],
: [http://dx.doi.org/10.1007/s00778-019-00547-y http://dx.doi.org/10.1007/s00778-019-00547-y] under the license https://creativecommons.org/licenses/by/4.0

* [https://dblp.uni-trier.de/db/journals/vldb/vldb29.html#LangPKBNK20 https://dblp.uni-trier.de/db/journals/vldb/vldb29.html#LangPKBNK20],
: [https://link.springer.com/article/10.1007/s00778-019-00547-y https://link.springer.com/article/10.1007/s00778-019-00547-y],
: [https://ir.cwi.nl/pub/28788 https://ir.cwi.nl/pub/28788],
: [https://ir.cwi.nl/pub/28788/Lang2019_Article_MakeTheMostOutOfYourSIMDInvest.pdf https://ir.cwi.nl/pub/28788/Lang2019_Article_MakeTheMostOutOfYourSIMDInvest.pdf],
: [https://academic.microsoft.com/#/detail/2961966181 https://academic.microsoft.com/#/detail/2961966181]

* [https://ir.cwi.nl/pub/27874 https://ir.cwi.nl/pub/27874],
: [http://dx.doi.org/10.1145/3211922.3211928 http://dx.doi.org/10.1145/3211922.3211928]

* [https://ir.cwi.nl/pub/28788 https://ir.cwi.nl/pub/28788],
: [http://dx.doi.org/10.1007/s00778-019-00547-y http://dx.doi.org/10.1007/s00778-019-00547-y]

* [https://www.narcis.nl/publication/RecordID/oai%3Acwi.nl%3A28788 https://www.narcis.nl/publication/RecordID/oai%3Acwi.nl%3A28788],
: [https://dblp.uni-trier.de/db/conf/damon/damon2018.html#LangKPB0K18 https://dblp.uni-trier.de/db/conf/damon/damon2018.html#LangKPB0K18],
: [https://ir.cwi.nl/pub/27874 https://ir.cwi.nl/pub/27874],
: [https://dl.acm.org/citation.cfm?doid=3211922.3211928 https://dl.acm.org/citation.cfm?doid=3211922.3211928],
: [https://ir.cwi.nl/pub/27874/27874.pdf https://ir.cwi.nl/pub/27874/27874.pdf],
: [https://dl.acm.org/citation.cfm?id=3211928 https://dl.acm.org/citation.cfm?id=3211928],
: [https://academic.microsoft.com/#/detail/2806132641 https://academic.microsoft.com/#/detail/2806132641]

DOIS: 10.1007/s00778-019-00547-y 10.1145/3211922.3211928

← Older revision	Revision as of 16:48, 28 January 2021
(No difference)

Passing et al 2019a - Revision history

Scipediacontent: Scipediacontent moved page Draft Content 333643256 to Passing et al 2019a

Scipediacontent: Created page with " == Abstract == Increasing single instruction multiple data (SIMD) capabilities in modern hardware allows for the compilation of data-parallel query pipelines. This means GPU..."