Active pebbles: Parallel programming for data-driven applications

Active pebbles: Parallel programming for data-driven applications. Willcock, J., J., Hoefler, T., Edmonds, N., G., & Lumsdaine, A. In Proceedings of the International Conference on Supercomputing, pages 235-244, 2011. ACM.

Website doi abstract bibtex

The scope of scientific computing continues to grow and now includes diverse application areas such as network analysis, combinatorialcomputing, and knowledge discovery, to name just a few. Large problems in these application areas require HPC resources, but they exhibit computation and communication patterns that are irregular, fine-grained, and non-local, making it difficult to apply traditional HPC approaches to achieve scalable solutions. In this paper we present Active Pebbles, a programming and execution model developed explicitly to enable the development of scalable software for these emerging application areas. Our approach relies on five main techniques - scalable addressing, active routing, message coalescing, message reduction, and termination detection - to separate algorithm expression from communication optimization. Using this approach, algorithms can be expressed in their natural forms, with their natural levels of granularity, while optimizations necessary for scalability can be applied automatically to match the characteristics of particular machines. We implement several example kernels using both Active Pebbles and existing programming models, evaluating both programmability and performance. Our experimental results demonstrate that the Active Pebbles model can succinctly and directly express irregular application kernels, while still achieving performance comparable to MPI-based implementations that are significantly more complex. © 2011 ACM.

@inproceedings{
 title = {Active pebbles: Parallel programming for data-driven applications},
 type = {inproceedings},
 year = {2011},
 keywords = {Active message,Active routing,Algorithms,Application area,Computer systems programming,Electric network analysis,Flocculati},
 pages = {235-244},
 websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-79959600862&doi=10.1145%2F1995896.1995934&partnerID=40&md5=e6c07eb3ac5e5e7f3ad670bbb69d502b},
 publisher = {ACM},
 city = {Tucson, AZ},
 id = {20ae1ecc-f9e2-33a9-9c90-dbf1b13c0933},
 created = {2017-11-27T19:30:57.292Z},
 file_attached = {false},
 profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},
 last_modified = {2018-03-12T19:03:32.189Z},
 read = {false},
 starred = {false},
 authored = {true},
 confirmed = {true},
 hidden = {false},
 citation_key = {Willcock2011235},
 source_type = {conference},
 notes = {<b>From Duplicate 1 (<i>Active pebbles: Parallel programming for data-driven applications</i> - Willcock, J J; Hoefler, T; Edmonds, N G; Lumsdaine, A)<br/></b><br/>cited By 28; Conference of 25th ACM International Conference on Supercomputing, ICS 2011 ; Conference Date: 31 May 2011 Through 4 June 2011; Conference Code:85314},
 folder_uuids = {a0f5ac31-a393-4a7b-b7db-64a126a80f6e},
 private_publication = {false},
 abstract = {The scope of scientific computing continues to grow and now includes diverse application areas such as network analysis, combinatorialcomputing, and knowledge discovery, to name just a few. Large problems in these application areas require HPC resources, but they exhibit computation and communication patterns that are irregular, fine-grained, and non-local, making it difficult to apply traditional HPC approaches to achieve scalable solutions. In this paper we present Active Pebbles, a programming and execution model developed explicitly to enable the development of scalable software for these emerging application areas. Our approach relies on five main techniques - scalable addressing, active routing, message coalescing, message reduction, and termination detection - to separate algorithm expression from communication optimization. Using this approach, algorithms can be expressed in their natural forms, with their natural levels of granularity, while optimizations necessary for scalability can be applied automatically to match the characteristics of particular machines. We implement several example kernels using both Active Pebbles and existing programming models, evaluating both programmability and performance. Our experimental results demonstrate that the Active Pebbles model can succinctly and directly express irregular application kernels, while still achieving performance comparable to MPI-based implementations that are significantly more complex. © 2011 ACM.},
 bibtype = {inproceedings},
 author = {Willcock, Jeremiah James and Hoefler, Torsten and Edmonds, Nicholas Gerard and Lumsdaine, Andrew},
 doi = {10.1145/1995896.1995934},
 booktitle = {Proceedings of the International Conference on Supercomputing}
}

Downloads: 0

{"_id":"F4z8Wj6hnrZ2P5q8u","bibbaseid":"willcock-hoefler-edmonds-lumsdaine-activepebblesparallelprogrammingfordatadrivenapplications-2011","downloads":0,"creationDate":"2018-03-12T19:10:27.405Z","title":"Active pebbles: Parallel programming for data-driven applications","author_short":["Willcock, J., J.","Hoefler, T.","Edmonds, N., G.","Lumsdaine, A."],"year":2011,"bibtype":"inproceedings","biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibdata":{"title":"Active pebbles: Parallel programming for data-driven applications","type":"inproceedings","year":"2011","keywords":"Active message,Active routing,Algorithms,Application area,Computer systems programming,Electric network analysis,Flocculati","pages":"235-244","websites":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-79959600862&doi=10.1145%2F1995896.1995934&partnerID=40&md5=e6c07eb3ac5e5e7f3ad670bbb69d502b","publisher":"ACM","city":"Tucson, AZ","id":"20ae1ecc-f9e2-33a9-9c90-dbf1b13c0933","created":"2017-11-27T19:30:57.292Z","file_attached":false,"profile_id":"42d295c0-0737-38d6-8b43-508cab6ea85d","last_modified":"2018-03-12T19:03:32.189Z","read":false,"starred":false,"authored":"true","confirmed":"true","hidden":false,"citation_key":"Willcock2011235","source_type":"conference","notes":"From Duplicate 1 (Active pebbles: Parallel programming for data-driven applications - Willcock, J J; Hoefler, T; Edmonds, N G; Lumsdaine, A) cited By 28; Conference of 25th ACM International Conference on Supercomputing, ICS 2011 ; Conference Date: 31 May 2011 Through 4 June 2011; Conference Code:85314","folder_uuids":"a0f5ac31-a393-4a7b-b7db-64a126a80f6e","private_publication":false,"abstract":"The scope of scientific computing continues to grow and now includes diverse application areas such as network analysis, combinatorialcomputing, and knowledge discovery, to name just a few. Large problems in these application areas require HPC resources, but they exhibit computation and communication patterns that are irregular, fine-grained, and non-local, making it difficult to apply traditional HPC approaches to achieve scalable solutions. In this paper we present Active Pebbles, a programming and execution model developed explicitly to enable the development of scalable software for these emerging application areas. Our approach relies on five main techniques - scalable addressing, active routing, message coalescing, message reduction, and termination detection - to separate algorithm expression from communication optimization. Using this approach, algorithms can be expressed in their natural forms, with their natural levels of granularity, while optimizations necessary for scalability can be applied automatically to match the characteristics of particular machines. We implement several example kernels using both Active Pebbles and existing programming models, evaluating both programmability and performance. Our experimental results demonstrate that the Active Pebbles model can succinctly and directly express irregular application kernels, while still achieving performance comparable to MPI-based implementations that are significantly more complex. © 2011 ACM.","bibtype":"inproceedings","author":"Willcock, Jeremiah James and Hoefler, Torsten and Edmonds, Nicholas Gerard and Lumsdaine, Andrew","doi":"10.1145/1995896.1995934","booktitle":"Proceedings of the International Conference on Supercomputing","bibtex":"@inproceedings{\n title = {Active pebbles: Parallel programming for data-driven applications},\n type = {inproceedings},\n year = {2011},\n keywords = {Active message,Active routing,Algorithms,Application area,Computer systems programming,Electric network analysis,Flocculati},\n pages = {235-244},\n websites = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-79959600862&doi=10.1145%2F1995896.1995934&partnerID=40&md5=e6c07eb3ac5e5e7f3ad670bbb69d502b},\n publisher = {ACM},\n city = {Tucson, AZ},\n id = {20ae1ecc-f9e2-33a9-9c90-dbf1b13c0933},\n created = {2017-11-27T19:30:57.292Z},\n file_attached = {false},\n profile_id = {42d295c0-0737-38d6-8b43-508cab6ea85d},\n last_modified = {2018-03-12T19:03:32.189Z},\n read = {false},\n starred = {false},\n authored = {true},\n confirmed = {true},\n hidden = {false},\n citation_key = {Willcock2011235},\n source_type = {conference},\n notes = {From Duplicate 1 (Active pebbles: Parallel programming for data-driven applications - Willcock, J J; Hoefler, T; Edmonds, N G; Lumsdaine, A) cited By 28; Conference of 25th ACM International Conference on Supercomputing, ICS 2011 ; Conference Date: 31 May 2011 Through 4 June 2011; Conference Code:85314},\n folder_uuids = {a0f5ac31-a393-4a7b-b7db-64a126a80f6e},\n private_publication = {false},\n abstract = {The scope of scientific computing continues to grow and now includes diverse application areas such as network analysis, combinatorialcomputing, and knowledge discovery, to name just a few. Large problems in these application areas require HPC resources, but they exhibit computation and communication patterns that are irregular, fine-grained, and non-local, making it difficult to apply traditional HPC approaches to achieve scalable solutions. In this paper we present Active Pebbles, a programming and execution model developed explicitly to enable the development of scalable software for these emerging application areas. Our approach relies on five main techniques - scalable addressing, active routing, message coalescing, message reduction, and termination detection - to separate algorithm expression from communication optimization. Using this approach, algorithms can be expressed in their natural forms, with their natural levels of granularity, while optimizations necessary for scalability can be applied automatically to match the characteristics of particular machines. We implement several example kernels using both Active Pebbles and existing programming models, evaluating both programmability and performance. Our experimental results demonstrate that the Active Pebbles model can succinctly and directly express irregular application kernels, while still achieving performance comparable to MPI-based implementations that are significantly more complex. © 2011 ACM.},\n bibtype = {inproceedings},\n author = {Willcock, Jeremiah James and Hoefler, Torsten and Edmonds, Nicholas Gerard and Lumsdaine, Andrew},\n doi = {10.1145/1995896.1995934},\n booktitle = {Proceedings of the International Conference on Supercomputing}\n}","author_short":["Willcock, J., J.","Hoefler, T.","Edmonds, N., G.","Lumsdaine, A."],"urls":{"Website":"https://www.scopus.com/inward/record.uri?eid=2-s2.0-79959600862&doi=10.1145%2F1995896.1995934&partnerID=40&md5=e6c07eb3ac5e5e7f3ad670bbb69d502b"},"biburl":"https://bibbase.org/service/mendeley/42d295c0-0737-38d6-8b43-508cab6ea85d","bibbaseid":"willcock-hoefler-edmonds-lumsdaine-activepebblesparallelprogrammingfordatadrivenapplications-2011","role":"author","keyword":["Active message","Active routing","Algorithms","Application area","Computer systems programming","Electric network analysis","Flocculati"],"metadata":{"authorlinks":{}},"downloads":0},"search_terms":["active","pebbles","parallel","programming","data","driven","applications","willcock","hoefler","edmonds","lumsdaine"],"keywords":["active message","active routing","algorithms","application area","computer systems programming","electric network analysis","flocculati"],"authorIDs":[],"dataSources":["zgahneP4uAjKbudrQ","ya2CyA73rpZseyrZ8","2252seNhipfTmjEBQ"]}