A high performance parallel algorithm for 1-D FFT
In this paper we propose a parallel high performance FFT algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of...
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Tagungsbericht |
| Sprache: | eng |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 40 |
|---|---|
| container_issue | |
| container_start_page | 34 |
| container_title | |
| container_volume | |
| creator | Agarwal, R C Gustavson, F G Zubair, M |
| description | In this paper we propose a parallel high performance FFT algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of an input sequence, multiplication of the transformed data by a coefficient array, and finally an inverse FFT computation of the resultant data. We show that the multidimensional formulation helps in reducing the communication costs and also improves the single node performance by effectively utilizing the memory system of the node. We implemented this kernel on the IBM SP1 and observed a performance of 1.25 GFLOPS on a 64-node machine. |
| doi_str_mv | 10.1145/602770.602784 |
| format | Conference Proceeding |
| fullrecord | <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_miscellaneous_31226821</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>31226821</sourcerecordid><originalsourceid>FETCH-LOGICAL-p1384-99751da8d6a7ed40474478333b2c74e0d6283a026cde05555497fb3be370f74f3</originalsourceid><addsrcrecordid>eNotjD1PwzAUAC1BJfrByO6JLeXZz_ZzxqoQQKrEUubKSV6aIKcJcfv_KYJbbjjphHhQsFbK2CcHmgjWv_LmRizAK--cA-tuxVyBwyy3aO_EIqUvAKTc0VyojWy7YytHnpph6sOpYjmGKcTIUYZ4HKbu3Pby2qTKnmVR7Fdi1oSY-P7fS_FZvOy3b9nu4_V9u9llo0Jvsjwnq-rgaxeIawOGjCGPiKWuyDDUTnsMoF1VM9grJqemxJKRoCHT4FI8_n3Hafi-cDof-i5VHGM48XBJB1RaO68V_gAxJkSE</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype><pqid>31226821</pqid></control><display><type>conference_proceeding</type><title>A high performance parallel algorithm for 1-D FFT</title><source>IEEE Electronic Library (IEL) Conference Proceedings</source><creator>Agarwal, R C ; Gustavson, F G ; Zubair, M</creator><creatorcontrib>Agarwal, R C ; Gustavson, F G ; Zubair, M</creatorcontrib><description>In this paper we propose a parallel high performance FFT algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of an input sequence, multiplication of the transformed data by a coefficient array, and finally an inverse FFT computation of the resultant data. We show that the multidimensional formulation helps in reducing the communication costs and also improves the single node performance by effectively utilizing the memory system of the node. We implemented this kernel on the IBM SP1 and observed a performance of 1.25 GFLOPS on a 64-node machine.</description><identifier>ISSN: 1063-9535</identifier><identifier>ISBN: 0818666056</identifier><identifier>ISBN: 9780818666056</identifier><identifier>DOI: 10.1145/602770.602784</identifier><language>eng</language><ispartof>Conference on High Performance Networking and Computing: Proceedings of the 1994 ACM/IEEE conference on Supercomputing : Washington, D.C.; 14-18 Nov. 1994, 1994, p.34-40</ispartof><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>309,310,780,784,789,790,27925</link.rule.ids></links><search><creatorcontrib>Agarwal, R C</creatorcontrib><creatorcontrib>Gustavson, F G</creatorcontrib><creatorcontrib>Zubair, M</creatorcontrib><title>A high performance parallel algorithm for 1-D FFT</title><title>Conference on High Performance Networking and Computing: Proceedings of the 1994 ACM/IEEE conference on Supercomputing : Washington, D.C.; 14-18 Nov. 1994</title><description>In this paper we propose a parallel high performance FFT algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of an input sequence, multiplication of the transformed data by a coefficient array, and finally an inverse FFT computation of the resultant data. We show that the multidimensional formulation helps in reducing the communication costs and also improves the single node performance by effectively utilizing the memory system of the node. We implemented this kernel on the IBM SP1 and observed a performance of 1.25 GFLOPS on a 64-node machine.</description><issn>1063-9535</issn><isbn>0818666056</isbn><isbn>9780818666056</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>1994</creationdate><recordtype>conference_proceeding</recordtype><recordid>eNotjD1PwzAUAC1BJfrByO6JLeXZz_ZzxqoQQKrEUubKSV6aIKcJcfv_KYJbbjjphHhQsFbK2CcHmgjWv_LmRizAK--cA-tuxVyBwyy3aO_EIqUvAKTc0VyojWy7YytHnpph6sOpYjmGKcTIUYZ4HKbu3Pby2qTKnmVR7Fdi1oSY-P7fS_FZvOy3b9nu4_V9u9llo0Jvsjwnq-rgaxeIawOGjCGPiKWuyDDUTnsMoF1VM9grJqemxJKRoCHT4FI8_n3Hafi-cDof-i5VHGM48XBJB1RaO68V_gAxJkSE</recordid><startdate>19941114</startdate><enddate>19941114</enddate><creator>Agarwal, R C</creator><creator>Gustavson, F G</creator><creator>Zubair, M</creator><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>19941114</creationdate><title>A high performance parallel algorithm for 1-D FFT</title><author>Agarwal, R C ; Gustavson, F G ; Zubair, M</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-p1384-99751da8d6a7ed40474478333b2c74e0d6283a026cde05555497fb3be370f74f3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>1994</creationdate><toplevel>online_resources</toplevel><creatorcontrib>Agarwal, R C</creatorcontrib><creatorcontrib>Gustavson, F G</creatorcontrib><creatorcontrib>Zubair, M</creatorcontrib><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Agarwal, R C</au><au>Gustavson, F G</au><au>Zubair, M</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>A high performance parallel algorithm for 1-D FFT</atitle><btitle>Conference on High Performance Networking and Computing: Proceedings of the 1994 ACM/IEEE conference on Supercomputing : Washington, D.C.; 14-18 Nov. 1994</btitle><date>1994-11-14</date><risdate>1994</risdate><spage>34</spage><epage>40</epage><pages>34-40</pages><issn>1063-9535</issn><isbn>0818666056</isbn><isbn>9780818666056</isbn><abstract>In this paper we propose a parallel high performance FFT algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of an input sequence, multiplication of the transformed data by a coefficient array, and finally an inverse FFT computation of the resultant data. We show that the multidimensional formulation helps in reducing the communication costs and also improves the single node performance by effectively utilizing the memory system of the node. We implemented this kernel on the IBM SP1 and observed a performance of 1.25 GFLOPS on a 64-node machine.</abstract><doi>10.1145/602770.602784</doi><tpages>7</tpages><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 1063-9535 |
| ispartof | Conference on High Performance Networking and Computing: Proceedings of the 1994 ACM/IEEE conference on Supercomputing : Washington, D.C.; 14-18 Nov. 1994, 1994, p.34-40 |
| issn | 1063-9535 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_31226821 |
| source | IEEE Electronic Library (IEL) Conference Proceedings |
| title | A high performance parallel algorithm for 1-D FFT |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-27T10%3A12%3A02IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=A%20high%20performance%20parallel%20algorithm%20for%201-D%20FFT&rft.btitle=Conference%20on%20High%20Performance%20Networking%20and%20Computing:%20Proceedings%20of%20the%201994%20ACM/IEEE%20conference%20on%20Supercomputing%20:%20Washington,%20D.C.;%2014-18%20Nov.%201994&rft.au=Agarwal,%20R%20C&rft.date=1994-11-14&rft.spage=34&rft.epage=40&rft.pages=34-40&rft.issn=1063-9535&rft.isbn=0818666056&rft.isbn_list=9780818666056&rft_id=info:doi/10.1145/602770.602784&rft_dat=%3Cproquest%3E31226821%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=31226821&rft_id=info:pmid/&rfr_iscdi=true |