FM-Index on GPU: A Cooperative Scheme to Reduce Memory Footprint

Alejandro Chacon; Santiago Marco; Antonio Espinosa; Paolo Ribeca; Juan Carlos Moure Lopez

doi:10.1109/ISPA.2014.10

FM-Index on GPU: A Cooperative Scheme to Reduce Memory Footprint

Alejandro Chacon, Santiago Marco, Antonio Espinosa, Paolo Ribeca, Juan Carlos Moure Lopez

Research output: Chapter in Book › Chapter › Research

8 Citations (Scopus)

Abstract

The FM-index is a data structure which is seeing more and more pervasive use, in particular in the field of high-throughput bioinformatics. Algorithms based on it show a pseudo-random memory access pattern. As a consequence, they are usually bound by memory bandwidth rather than CPU usage. Naive GPU implementations are no exception. Here we show that the combination of a compact design of the FM-index and a thread-cooperative approach can be used to restore a proper balance. The resulting solution is less memory-bandwidth intensive, and allows full exploitation of the computational resources of the GPU across several GPU architectures.

Original language	English
Title of host publication	Parallel and Distributed Processing with Applications (ISPA), 2014 IEEE International Symposium on
Place of Publication	Milà (IT)
Publisher	IEEE Computer Society
Pages	1-9
Number of pages	8
Volume	1
Edition	1
ISBN (Print)	978-1-4799-4293-0
DOIs	https://doi.org/10.1109/ISPA.2014.10
Publication status	Published - 1 Jan 2014

Access to Document

10.1109/ISPA.2014.10

https://ddd.uab.cat/record/131866

Cite this

@inbook{471ab7edaf954474bf55b618854b1802,

title = "FM-Index on GPU: A Cooperative Scheme to Reduce Memory Footprint",

abstract = "The FM-index is a data structure which is seeing more and more pervasive use, in particular in the field of high-throughput bioinformatics. Algorithms based on it show a pseudo-random memory access pattern. As a consequence, they are usually bound by memory bandwidth rather than CPU usage. Naive GPU implementations are no exception. Here we show that the combination of a compact design of the FM-index and a thread-cooperative approach can be used to restore a proper balance. The resulting solution is less memory-bandwidth intensive, and allows full exploitation of the computational resources of the GPU across several GPU architectures.",

author = "Alejandro Chacon and Santiago Marco and Antonio Espinosa and Paolo Ribeca and \{Moure Lopez\}, \{Juan Carlos\}",

year = "2014",

month = jan,

day = "1",

doi = "10.1109/ISPA.2014.10",

language = "English",

isbn = "978-1-4799-4293-0",

volume = "1",

pages = "1--9",

booktitle = "Parallel and Distributed Processing with Applications (ISPA), 2014 IEEE International Symposium on",

publisher = "IEEE Computer Society",

address = "United States",

edition = "1",

}

TY - CHAP

T1 - FM-Index on GPU: A Cooperative Scheme to Reduce Memory Footprint

AU - Chacon, Alejandro

AU - Marco, Santiago

AU - Espinosa, Antonio

AU - Ribeca, Paolo

AU - Moure Lopez, Juan Carlos

PY - 2014/1/1

Y1 - 2014/1/1

N2 - The FM-index is a data structure which is seeing more and more pervasive use, in particular in the field of high-throughput bioinformatics. Algorithms based on it show a pseudo-random memory access pattern. As a consequence, they are usually bound by memory bandwidth rather than CPU usage. Naive GPU implementations are no exception. Here we show that the combination of a compact design of the FM-index and a thread-cooperative approach can be used to restore a proper balance. The resulting solution is less memory-bandwidth intensive, and allows full exploitation of the computational resources of the GPU across several GPU architectures.

AB - The FM-index is a data structure which is seeing more and more pervasive use, in particular in the field of high-throughput bioinformatics. Algorithms based on it show a pseudo-random memory access pattern. As a consequence, they are usually bound by memory bandwidth rather than CPU usage. Naive GPU implementations are no exception. Here we show that the combination of a compact design of the FM-index and a thread-cooperative approach can be used to restore a proper balance. The resulting solution is less memory-bandwidth intensive, and allows full exploitation of the computational resources of the GPU across several GPU architectures.

UR - http://www.scopus.com/inward/record.url?eid=2-s2.0-84911439685&partnerID=MN8TOARS

UR - https://www.scopus.com/pages/publications/84911439685

U2 - 10.1109/ISPA.2014.10

DO - 10.1109/ISPA.2014.10

M3 - Chapter

SN - 978-1-4799-4293-0

VL - 1

SP - 1

EP - 9

BT - Parallel and Distributed Processing with Applications (ISPA), 2014 IEEE International Symposium on

PB - IEEE Computer Society

CY - Milà (IT)

ER -

FM-Index on GPU: A Cooperative Scheme to Reduce Memory Footprint

Abstract

Access to Document

Other files and links

Fingerprint

Cite this