Accurately measuring collective operations at massive scale. Hoefler, T., Schneider, T., & Lumsdaine, A. In IPDPS Miami 2008 - Proceedings of the 22nd IEEE International Parallel and Distributed Processing Symposium, Program and CD-ROM, 2008. ![Accurately measuring collective operations at massive scale [link]](https://bibbase.org/img/filetypes/link.svg "link")
Website doi abstract bibtex Accurate, reproducible and comparable measurement of collective operations is a complicated task. Although Different measurement schemes are implemented in wellknown benchmarks, many of these schemes introduce different systematic errors in their measurements. We characterize these errors and select a window-based approach as the most accurate method. However, this approach complicates measurements significantly and introduces a clock synchronization as a new source of systematic errors. We analyze approaches to avoid or correct those errors and develop a scalable synchronization scheme to conduct benchmarks on massively parallel systems. Our results are compared to the window-based scheme implemented in the SKaMPI benchmarks and show a reduction of the synchronization overhead by a factor of 16 on 128 processes. ©2008 IEEE.
@inproceedings{
title = {Accurately measuring collective operations at massive scale},
type = {inproceedings},
year = {2008},
keywords = {Benchmarking; Collective operations; MPI; Scalabl,Computer networks; Distributed parameter networks;,Measurement errors},
websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-51049102790&doi=10.1109%2FIPDPS.2008.4536494&partnerID=40&md5=89d80c9f9a6890223995795f519e5567},
city = {Miami, FL},
id = {3d8aa8b8-666c-3420-be59-1c5d2eaed4ca},
created = {2018-01-09T20:30:38.343Z},
file_attached = {false},
profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
last_modified = {2018-03-12T19:03:18.401Z},
read = {false},
starred = {false},
authored = {true},
confirmed = {true},
hidden = {false},
citation_key = {Hoefler2008},
source_type = {conference},
notes = {cited By 21; Conference of IPDPS 2008 - 22nd IEEE International Parallel and Distributed Processing Symposium ; Conference Date: 14 April 2008 Through 18 April 2008; Conference Code:73339},
folder_uuids = {2aba6c14-9027-4f47-8627-0902e1e2342b},
private_publication = {false},
abstract = {Accurate, reproducible and comparable measurement of collective operations is a complicated task. Although Different measurement schemes are implemented in wellknown benchmarks, many of these schemes introduce different systematic errors in their measurements. We characterize these errors and select a window-based approach as the most accurate method. However, this approach complicates measurements significantly and introduces a clock synchronization as a new source of systematic errors. We analyze approaches to avoid or correct those errors and develop a scalable synchronization scheme to conduct benchmarks on massively parallel systems. Our results are compared to the window-based scheme implemented in the SKaMPI benchmarks and show a reduction of the synchronization overhead by a factor of 16 on 128 processes. ©2008 IEEE.},
bibtype = {inproceedings},
author = {Hoefler, T and Schneider, T and Lumsdaine, A},
doi = {10.1109/IPDPS.2008.4536494},
booktitle = {IPDPS Miami 2008 - Proceedings of the 22nd IEEE International Parallel and Distributed Processing Symposium, Program and CD-ROM}
}
Downloads: 0
{"_id":"eopmKghozJDga99yr","bibbaseid":"hoefler-schneider-lumsdaine-accuratelymeasuringcollectiveoperationsatmassivescale-2008","downloads":0,"creationDate":"2018-03-12T19:10:28.099Z","title":"Accurately measuring collective operations at massive scale","author_short":["Hoefler, T.","Schneider, T.","Lumsdaine, A."],"year":2008,"bibtype":"inproceedings","biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibdata":{"title":"Accurately measuring collective operations at massive scale","type":"inproceedings","year":"2008","keywords":"Benchmarking; Collective operations; MPI; Scalabl,Computer networks; Distributed parameter networks;,Measurement errors","websites":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-51049102790&doi=10.1109%2FIPDPS.2008.4536494&partnerID=40&md5=89d80c9f9a6890223995795f519e5567","city":"Miami, FL","id":"3d8aa8b8-666c-3420-be59-1c5d2eaed4ca","created":"2018-01-09T20:30:38.343Z","file_attached":false,"profile_id":"42d295c0-0737-38d6-8b43-508cab6ea85d","last_modified":"2018-03-12T19:03:18.401Z","read":false,"starred":false,"authored":"true","confirmed":"true","hidden":false,"citation_key":"Hoefler2008","source_type":"conference","notes":"cited By 21; Conference of IPDPS 2008 - 22nd IEEE International Parallel and Distributed Processing Symposium ; Conference Date: 14 April 2008 Through 18 April 2008; Conference Code:73339","folder_uuids":"2aba6c14-9027-4f47-8627-0902e1e2342b","private_publication":false,"abstract":"Accurate, reproducible and comparable measurement of collective operations is a complicated task. Although Different measurement schemes are implemented in wellknown benchmarks, many of these schemes introduce different systematic errors in their measurements. We characterize these errors and select a window-based approach as the most accurate method. However, this approach complicates measurements significantly and introduces a clock synchronization as a new source of systematic errors. We analyze approaches to avoid or correct those errors and develop a scalable synchronization scheme to conduct benchmarks on massively parallel systems. Our results are compared to the window-based scheme implemented in the SKaMPI benchmarks and show a reduction of the synchronization overhead by a factor of 16 on 128 processes. ©2008 IEEE.","bibtype":"inproceedings","author":"Hoefler, T and Schneider, T and Lumsdaine, A","doi":"10.1109/IPDPS.2008.4536494","booktitle":"IPDPS Miami 2008 - Proceedings of the 22nd IEEE International Parallel and Distributed Processing Symposium, Program and CD-ROM","bibtex":"@inproceedings{\n title = {Accurately measuring collective operations at massive scale},\n type = {inproceedings},\n year = {2008},\n keywords = {Benchmarking; Collective operations; MPI; Scalabl,Computer networks; Distributed parameter networks;,Measurement errors},\n websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-51049102790&doi=10.1109%2FIPDPS.2008.4536494&partnerID=40&md5=89d80c9f9a6890223995795f519e5567},\n city = {Miami, FL},\n id = {3d8aa8b8-666c-3420-be59-1c5d2eaed4ca},\n created = {2018-01-09T20:30:38.343Z},\n file_attached = {false},\n profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},\n last_modified = {2018-03-12T19:03:18.401Z},\n read = {false},\n starred = {false},\n authored = {true},\n confirmed = {true},\n hidden = {false},\n citation_key = {Hoefler2008},\n source_type = {conference},\n notes = {cited By 21; Conference of IPDPS 2008 - 22nd IEEE International Parallel and Distributed Processing Symposium ; Conference Date: 14 April 2008 Through 18 April 2008; Conference Code:73339},\n folder_uuids = {2aba6c14-9027-4f47-8627-0902e1e2342b},\n private_publication = {false},\n abstract = {Accurate, reproducible and comparable measurement of collective operations is a complicated task. Although Different measurement schemes are implemented in wellknown benchmarks, many of these schemes introduce different systematic errors in their measurements. We characterize these errors and select a window-based approach as the most accurate method. However, this approach complicates measurements significantly and introduces a clock synchronization as a new source of systematic errors. We analyze approaches to avoid or correct those errors and develop a scalable synchronization scheme to conduct benchmarks on massively parallel systems. Our results are compared to the window-based scheme implemented in the SKaMPI benchmarks and show a reduction of the synchronization overhead by a factor of 16 on 128 processes. ©2008 IEEE.},\n bibtype = {inproceedings},\n author = {Hoefler, T and Schneider, T and Lumsdaine, A},\n doi = {10.1109/IPDPS.2008.4536494},\n booktitle = {IPDPS Miami 2008 - Proceedings of the 22nd IEEE International Parallel and Distributed Processing Symposium, Program and CD-ROM}\n}","author_short":["Hoefler, T.","Schneider, T.","Lumsdaine, A."],"urls":{"Website":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-51049102790&doi=10.1109%2FIPDPS.2008.4536494&partnerID=40&md5=89d80c9f9a6890223995795f519e5567"},"biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibbaseid":"hoefler-schneider-lumsdaine-accuratelymeasuringcollectiveoperationsatmassivescale-2008","role":"author","keyword":["Benchmarking; Collective operations; MPI; Scalabl","Computer networks; Distributed parameter networks;","Measurement errors"],"metadata":{"authorlinks":{}},"downloads":0},"search_terms":["accurately","measuring","collective","operations","massive","scale","hoefler","schneider","lumsdaine"],"keywords":["benchmarking; collective operations; mpi; scalabl","computer networks; distributed parameter networks;","measurement errors"],"authorIDs":[],"dataSources":["zgahneP4uAjKbudrQ","ya2CyA73rpZseyrZ8"]}