LEADER 02066nam a2200289 c 4500
001 22-14-qucosa2-766710
007 cr
008 2021 eng
037 |a urn:nbn:de:bsz:14-qucosa2-766710 
041 |a eng 
082 |a 621.3 
100 |a Zarubin, Mikhail 
245 |a SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred 
264 |a New York  |b ACM  |c 2021 
336 |b txt 
338 |b nc 
533 |a Online-Ausg.  |d 2021  |e Online-Ressource (Text)  |f Technische Universität Dresden 
520 |a Hybrid memory systems consisting of DRAM and NVRAM offer a great opportunity for column-oriented data systems to persistently store and to efficiently process columnar data completely in main memory. While vectorization (SIMD) of query operators is state-of-the-art to increase the single-thread performance, it has to be combined with thread-level parallelism (MIMD) to satisfy growing needs for higher performance and scalability. However, it is not well investigated how such a SIMD-MIMD interplay could be leveraged efficiently in hybrid memory systems. On the one hand, we deliver an extensive experimental evaluation of typical workloads on columnar data in this paper. We reveal that the choice of the most performant SIMD version differs greatly for both memory types. Moreover, we show that the throughput of concurrent queries can be boosted (up to 2x) when combining various SIMD flavors in a multi-threaded execution. On the other hand, to enable that optimization, we propose an adaptive SIMD-MIMD cocktail approach incurring only a negligible runtime overhead. 
650 |a Hybrid Memory; Column Store; Simd; Mimd; Optimization 
650 |a Hybridspeicher; Spaltenorientierter Speicher; Simd; Mimd; Optimierung 
655 |a Konferenzschrift 
700 |a Damme, Patrick 
700 |a Krause, Alexander 
700 |a Habich, Dirk 
700 |a Lehner, Wolfgang 
856 4 0 |q text/html  |u https://nbn-resolving.org/urn:nbn:de:bsz:14-qucosa2-766710  |z Online-Zugriff 
980 |a 14-qucosa2-766710  |b 22  |c sid-22-col-qucosa 
openURL url_ver=Z39.88-2004&ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fvufind.svn.sourceforge.net%3Agenerator&rft.title=SIMD-MIMD+cocktail+in+a+hybrid+memory+glass%3A+shaken%2C+not+stirred&rft.date=2021&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.creator=Zarubin%2C+Mikhail&rft.pub=ACM&rft.format=eBook&rft.language=English
SOLR
_version_ 1794846702356660224
author Zarubin, Mikhail
author2 Damme, Patrick, Krause, Alexander, Habich, Dirk, Lehner, Wolfgang
author2_role , , ,
author2_variant p d pd, a k ak, d h dh, w l wl
author_facet Zarubin, Mikhail, Damme, Patrick, Krause, Alexander, Habich, Dirk, Lehner, Wolfgang
author_role
author_sort Zarubin, Mikhail
author_variant m z mz
building Library A
collection sid-22-col-qucosa
contents Hybrid memory systems consisting of DRAM and NVRAM offer a great opportunity for column-oriented data systems to persistently store and to efficiently process columnar data completely in main memory. While vectorization (SIMD) of query operators is state-of-the-art to increase the single-thread performance, it has to be combined with thread-level parallelism (MIMD) to satisfy growing needs for higher performance and scalability. However, it is not well investigated how such a SIMD-MIMD interplay could be leveraged efficiently in hybrid memory systems. On the one hand, we deliver an extensive experimental evaluation of typical workloads on columnar data in this paper. We reveal that the choice of the most performant SIMD version differs greatly for both memory types. Moreover, we show that the throughput of concurrent queries can be boosted (up to 2x) when combining various SIMD flavors in a multi-threaded execution. On the other hand, to enable that optimization, we propose an adaptive SIMD-MIMD cocktail approach incurring only a negligible runtime overhead.
dewey-full 621.3
dewey-hundreds 600 - Technology (Applied sciences)
dewey-ones 621 - Applied physics
dewey-raw 621.3
dewey-search 621.3
dewey-sort 3621.3
dewey-tens 620 - Engineering and allied operations
facet_avail Online, Free
finc_class_facet Technik
fincclass_txtF_mv engineering-electrical, engineering-process, technology
format eBook
format_access_txtF_mv Book, E-Book
format_de105 Ebook
format_de14 Book, E-Book
format_de15 Book, E-Book
format_del152 Buch
format_detail_txtF_mv text-online-monograph-independent-conference
format_dezi4 e-Book
format_finc Book, E-Book
format_legacy ElectronicBook
format_legacy_nrw Book, E-Book
format_nrw Book, E-Book
format_strict_txtF_mv E-Book
genre Konferenzschrift
genre_facet Konferenzschrift
geogr_code not assigned
geogr_code_person not assigned
id 22-14-qucosa2-766710
illustrated Not Illustrated
imprint New York, ACM, 2021
imprint_str_mv Online-Ausg.: 2021
institution DE-105, DE-Gla1, DE-Brt1, DE-D161, DE-540, DE-Pl11, DE-Rs1, DE-Bn3, DE-Zi4, DE-Zwi2, DE-D117, DE-Mh31, DE-D275, DE-Ch1, DE-15, DE-D13, DE-L242, DE-L229, DE-L328
is_hierarchy_id
is_hierarchy_title
language English
last_indexed 2024-03-29T08:00:01.234Z
match_str zarubin2021simdmimdcocktailinahybridmemoryglassshakennotstirred
mega_collection Qucosa
publishDate 2021
publishDateSort 2021
publishPlace New York
publisher ACM
record_format marcfinc
record_id 14-qucosa2-766710
recordtype marcfinc
rvk_facet No subject assigned
score 17,085752
source_id 22
spelling Zarubin, Mikhail, SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred, New York ACM 2021, txt, nc, Online-Ausg. 2021 Online-Ressource (Text) Technische Universität Dresden, Hybrid memory systems consisting of DRAM and NVRAM offer a great opportunity for column-oriented data systems to persistently store and to efficiently process columnar data completely in main memory. While vectorization (SIMD) of query operators is state-of-the-art to increase the single-thread performance, it has to be combined with thread-level parallelism (MIMD) to satisfy growing needs for higher performance and scalability. However, it is not well investigated how such a SIMD-MIMD interplay could be leveraged efficiently in hybrid memory systems. On the one hand, we deliver an extensive experimental evaluation of typical workloads on columnar data in this paper. We reveal that the choice of the most performant SIMD version differs greatly for both memory types. Moreover, we show that the throughput of concurrent queries can be boosted (up to 2x) when combining various SIMD flavors in a multi-threaded execution. On the other hand, to enable that optimization, we propose an adaptive SIMD-MIMD cocktail approach incurring only a negligible runtime overhead., Hybrid Memory; Column Store; Simd; Mimd; Optimization, Hybridspeicher; Spaltenorientierter Speicher; Simd; Mimd; Optimierung, Konferenzschrift, Damme, Patrick, Krause, Alexander, Habich, Dirk, Lehner, Wolfgang, text/html https://nbn-resolving.org/urn:nbn:de:bsz:14-qucosa2-766710 Online-Zugriff
spellingShingle Zarubin, Mikhail, SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred, Hybrid memory systems consisting of DRAM and NVRAM offer a great opportunity for column-oriented data systems to persistently store and to efficiently process columnar data completely in main memory. While vectorization (SIMD) of query operators is state-of-the-art to increase the single-thread performance, it has to be combined with thread-level parallelism (MIMD) to satisfy growing needs for higher performance and scalability. However, it is not well investigated how such a SIMD-MIMD interplay could be leveraged efficiently in hybrid memory systems. On the one hand, we deliver an extensive experimental evaluation of typical workloads on columnar data in this paper. We reveal that the choice of the most performant SIMD version differs greatly for both memory types. Moreover, we show that the throughput of concurrent queries can be boosted (up to 2x) when combining various SIMD flavors in a multi-threaded execution. On the other hand, to enable that optimization, we propose an adaptive SIMD-MIMD cocktail approach incurring only a negligible runtime overhead., Hybrid Memory; Column Store; Simd; Mimd; Optimization, Hybridspeicher; Spaltenorientierter Speicher; Simd; Mimd; Optimierung, Konferenzschrift
title SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_auth SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_full SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_fullStr SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_full_unstemmed SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_short SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
title_sort simd-mimd cocktail in a hybrid memory glass: shaken, not stirred
title_unstemmed SIMD-MIMD cocktail in a hybrid memory glass: shaken, not stirred
topic Hybrid Memory; Column Store; Simd; Mimd; Optimization, Hybridspeicher; Spaltenorientierter Speicher; Simd; Mimd; Optimierung, Konferenzschrift
topic_facet Hybrid Memory; Column Store; Simd; Mimd; Optimization, Hybridspeicher; Spaltenorientierter Speicher; Simd; Mimd; Optimierung, Konferenzschrift
url https://nbn-resolving.org/urn:nbn:de:bsz:14-qucosa2-766710
urn urn:nbn:de:bsz:14-qucosa2-766710
work_keys_str_mv AT zarubinmikhail simdmimdcocktailinahybridmemoryglassshakennotstirred, AT dammepatrick simdmimdcocktailinahybridmemoryglassshakennotstirred, AT krausealexander simdmimdcocktailinahybridmemoryglassshakennotstirred, AT habichdirk simdmimdcocktailinahybridmemoryglassshakennotstirred, AT lehnerwolfgang simdmimdcocktailinahybridmemoryglassshakennotstirred