Brief announcement: Leveraging non-blocking collective communication in high-performance applications. Hoefler, T., Gottschling, P., & Lumsdain, A. In Annual ACM Symposium on Parallelism in Algorithms and Architectures, pages 113-115, 2008. Website doi abstract bibtex Although overlapping communication with computation is an important mechanism for achieving high performance in parallel programs, developing applications that actually achieve good overlap can be difficult. Existing approaches are typically based on manual or compiler-based transformations. This paper presents a pattern and library-based approach to optimizing collective communication in parallel high-performance applications, based on using non-blocking collective operations to enable overlapping of communication and computation. Common communication and computation patterns in iterative SPMD computations are used to motivate the transformations we present. Our approach provides the programmer with the capability to separately optimize communication and computation in an application, while automating the interaction between computation and communication to achieve maximum overlap. Performance results with a model application show more than a 90% decrease in communication overhead, resulting in 21% overall performance improvements.
@inproceedings{
title = {Brief announcement: Leveraging non-blocking collective communication in high-performance applications},
type = {inproceedings},
year = {2008},
keywords = {Applications; Fourier transforms; Large scale syst,Collective communications; Collective operations;,Communication},
pages = {113-115},
websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-57549114097&doi=10.1145%2F1378533.1378554&partnerID=40&md5=5ee9347c567ead51a63e5264b758e6e0},
city = {Munich},
id = {e2efe601-d2c9-3464-bf2c-b253b44ee239},
created = {2018-01-09T20:30:38.063Z},
file_attached = {false},
profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
last_modified = {2018-03-12T19:03:18.629Z},
read = {false},
starred = {false},
authored = {true},
confirmed = {true},
hidden = {false},
citation_key = {Hoefler2008113},
source_type = {conference},
notes = {cited By 18; Conference of 20th ACM Symposium on Parallelism in Algorithms and Architectures, SPAA'08 ; Conference Date: 14 June 2008 Through 16 June 2008; Conference Code:74450},
folder_uuids = {2aba6c14-9027-4f47-8627-0902e1e2342b},
private_publication = {false},
abstract = {Although overlapping communication with computation is an important mechanism for achieving high performance in parallel programs, developing applications that actually achieve good overlap can be difficult. Existing approaches are typically based on manual or compiler-based transformations. This paper presents a pattern and library-based approach to optimizing collective communication in parallel high-performance applications, based on using non-blocking collective operations to enable overlapping of communication and computation. Common communication and computation patterns in iterative SPMD computations are used to motivate the transformations we present. Our approach provides the programmer with the capability to separately optimize communication and computation in an application, while automating the interaction between computation and communication to achieve maximum overlap. Performance results with a model application show more than a 90% decrease in communication overhead, resulting in 21% overall performance improvements.},
bibtype = {inproceedings},
author = {Hoefler, T and Gottschling, P and Lumsdain, A},
doi = {10.1145/1378533.1378554},
booktitle = {Annual ACM Symposium on Parallelism in Algorithms and Architectures}
}
Downloads: 0
{"_id":"Th33RJQaw3Ncc2zRP","bibbaseid":"hoefler-gottschling-lumsdain-briefannouncementleveragingnonblockingcollectivecommunicationinhighperformanceapplications-2008","downloads":0,"creationDate":"2018-03-12T19:10:28.090Z","title":"Brief announcement: Leveraging non-blocking collective communication in high-performance applications","author_short":["Hoefler, T.","Gottschling, P.","Lumsdain, A."],"year":2008,"bibtype":"inproceedings","biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibdata":{"title":"Brief announcement: Leveraging non-blocking collective communication in high-performance applications","type":"inproceedings","year":"2008","keywords":"Applications; Fourier transforms; Large scale syst,Collective communications; Collective operations;,Communication","pages":"113-115","websites":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-57549114097&doi=10.1145%2F1378533.1378554&partnerID=40&md5=5ee9347c567ead51a63e5264b758e6e0","city":"Munich","id":"e2efe601-d2c9-3464-bf2c-b253b44ee239","created":"2018-01-09T20:30:38.063Z","file_attached":false,"profile_id":"42d295c0-0737-38d6-8b43-508cab6ea85d","last_modified":"2018-03-12T19:03:18.629Z","read":false,"starred":false,"authored":"true","confirmed":"true","hidden":false,"citation_key":"Hoefler2008113","source_type":"conference","notes":"cited By 18; Conference of 20th ACM Symposium on Parallelism in Algorithms and Architectures, SPAA'08 ; Conference Date: 14 June 2008 Through 16 June 2008; Conference Code:74450","folder_uuids":"2aba6c14-9027-4f47-8627-0902e1e2342b","private_publication":false,"abstract":"Although overlapping communication with computation is an important mechanism for achieving high performance in parallel programs, developing applications that actually achieve good overlap can be difficult. Existing approaches are typically based on manual or compiler-based transformations. This paper presents a pattern and library-based approach to optimizing collective communication in parallel high-performance applications, based on using non-blocking collective operations to enable overlapping of communication and computation. Common communication and computation patterns in iterative SPMD computations are used to motivate the transformations we present. Our approach provides the programmer with the capability to separately optimize communication and computation in an application, while automating the interaction between computation and communication to achieve maximum overlap. Performance results with a model application show more than a 90% decrease in communication overhead, resulting in 21% overall performance improvements.","bibtype":"inproceedings","author":"Hoefler, T and Gottschling, P and Lumsdain, A","doi":"10.1145/1378533.1378554","booktitle":"Annual ACM Symposium on Parallelism in Algorithms and Architectures","bibtex":"@inproceedings{\n title = {Brief announcement: Leveraging non-blocking collective communication in high-performance applications},\n type = {inproceedings},\n year = {2008},\n keywords = {Applications; Fourier transforms; Large scale syst,Collective communications; Collective operations;,Communication},\n pages = {113-115},\n websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-57549114097&doi=10.1145%2F1378533.1378554&partnerID=40&md5=5ee9347c567ead51a63e5264b758e6e0},\n city = {Munich},\n id = {e2efe601-d2c9-3464-bf2c-b253b44ee239},\n created = {2018-01-09T20:30:38.063Z},\n file_attached = {false},\n profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},\n last_modified = {2018-03-12T19:03:18.629Z},\n read = {false},\n starred = {false},\n authored = {true},\n confirmed = {true},\n hidden = {false},\n citation_key = {Hoefler2008113},\n source_type = {conference},\n notes = {cited By 18; Conference of 20th ACM Symposium on Parallelism in Algorithms and Architectures, SPAA'08 ; Conference Date: 14 June 2008 Through 16 June 2008; Conference Code:74450},\n folder_uuids = {2aba6c14-9027-4f47-8627-0902e1e2342b},\n private_publication = {false},\n abstract = {Although overlapping communication with computation is an important mechanism for achieving high performance in parallel programs, developing applications that actually achieve good overlap can be difficult. Existing approaches are typically based on manual or compiler-based transformations. This paper presents a pattern and library-based approach to optimizing collective communication in parallel high-performance applications, based on using non-blocking collective operations to enable overlapping of communication and computation. Common communication and computation patterns in iterative SPMD computations are used to motivate the transformations we present. Our approach provides the programmer with the capability to separately optimize communication and computation in an application, while automating the interaction between computation and communication to achieve maximum overlap. Performance results with a model application show more than a 90% decrease in communication overhead, resulting in 21% overall performance improvements.},\n bibtype = {inproceedings},\n author = {Hoefler, T and Gottschling, P and Lumsdain, A},\n doi = {10.1145/1378533.1378554},\n booktitle = {Annual ACM Symposium on Parallelism in Algorithms and Architectures}\n}","author_short":["Hoefler, T.","Gottschling, P.","Lumsdain, A."],"urls":{"Website":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-57549114097&doi=10.1145%2F1378533.1378554&partnerID=40&md5=5ee9347c567ead51a63e5264b758e6e0"},"biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibbaseid":"hoefler-gottschling-lumsdain-briefannouncementleveragingnonblockingcollectivecommunicationinhighperformanceapplications-2008","role":"author","keyword":["Applications; Fourier transforms; Large scale syst","Collective communications; Collective operations;","Communication"],"metadata":{"authorlinks":{}},"downloads":0},"search_terms":["brief","announcement","leveraging","non","blocking","collective","communication","high","performance","applications","hoefler","gottschling","lumsdain"],"keywords":["applications; fourier transforms; large scale syst","collective communications; collective operations;","communication"],"authorIDs":[],"dataSources":["zgahneP4uAjKbudrQ","ya2CyA73rpZseyrZ8","2252seNhipfTmjEBQ"]}