inria-00379168, version 1

Voir la fiche détaillée

BibTeX  EndNote  TEI  RefWorks %% inria-00379168, version 1 %% http://hal.inria.fr/inria-00379168/en/ @inproceedings{GOGLIN:2009:INRIA-00379168:1, title = { {NIC}-assisted {C}ache-{E}fficient {R}eceive {S}tack for {M}essage {P}assing over {E}thernet}, author = {{G}oglin, {B}rice}, abstract = {{H}igh-speed networking in clusters usually relies on advanced hardware features in the {NIC}s, such as zero-copy. {O}pen-{MX} is a high-performance message passing stack designed for regular {E}thernet hardware without such capabilities. {W}e present the addition of multiqueue support in the {O}pen-{MX} receive stack so that all incoming packets for the same process are treated on the same core. {W}e then introduce the idea of binding the target end process near its dedicated receive queue. {T}his model leads to a more cache-efficient receive stack for {O}pen-{MX}. {I}t also proves that very simple and stateless hardware features may have a significant impact on message passing performance over {E}thernet. {T}he implementation of this model in a firmware reveals that it may not be as efficient as some manually tuned micro-benchmarks. {B}ut our multiqueue receive stack generally performs better than the original single queue stack, especially on large communication patterns where multiple processes are involved and manual binding is difficult.}, language = {{A}nglais}, affiliation = {{L}aboratoire {B}ordelais de {R}echerche en {I}nformatique - {L}a{BRI} - {CNRS} : {UMR}5800 - {U}niversit{\'e} {S}ciences et {T}echnologies - {B}ordeaux {I} - {E}cole {N}ationale {S}up{\'e}rieure d'{E}lectronique, {I}nformatique et {R}adiocommunications de {B}ordeaux - {U}niversit{\'e} {V}ictor {S}egalen - {B}ordeaux {II} - {RUNTIME} - {INRIA} {B}ordeaux - {S}ud-{O}uest - {INRIA} - {CNRS} : {UMR}5800 - {U}niversit{\'e} {S}ciences et {T}echnologies - {B}ordeaux {I} - {E}cole {N}ationale {S}up{\'e}rieure d'{E}lectronique, {I}nformatique et {R}adiocommunications de {B}ordeaux }, booktitle = {15th {I}nternational {E}uro-{P}ar {C}onference {L}ecture {N}otes in {C}omputer {S}cience }, address = {{D}elft {P}ays-{B}as }, editor = {{S}pringer {V}erlag }, audience = {internationale }, year = {2009}, URL = {http://hal.inria.fr/inria-00379168/en/}, } %F inria-00379168, version 1 %0 Conference Proceedings %U http://hal.inria.fr/inria-00379168/en/ %2 INFO:INFO_OS %T NIC-assisted Cache-Efficient Receive Stack for Message Passing over Ethernet %A Goglin, Brice %A Goglin, B. %A Goglin B. %+ Laboratoire Bordelais de Recherche en Informatique - LaBRI - CNRS : UMR5800 - Université Sciences et Technologies - Bordeaux I - Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux - Université Victor Segalen - Bordeaux II - RUNTIME - INRIA Bordeaux - Sud-Ouest - INRIA - CNRS : UMR5800 - Université Sciences et Technologies - Bordeaux I - Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux %X High-speed networking in clusters usually relies on advanced hardware features in the NICs, such as zero-copy. Open-MX is a high-performance message passing stack designed for regular Ethernet hardware without such capabilities. We present the addition of multiqueue support in the Open-MX receive stack so that all incoming packets for the same process are treated on the same core. We then introduce the idea of binding the target end process near its dedicated receive queue. This model leads to a more cache-efficient receive stack for Open-MX. It also proves that very simple and stateless hardware features may have a significant impact on message passing performance over Ethernet. The implementation of this model in a firmware reveals that it may not be as efficient as some manually tuned micro-benchmarks. But our multiqueue receive stack generally performs better than the original single queue stack, especially on large communication patterns where multiple processes are involved and manual binding is difficult. %G Anglais %8 2009 %2 internationale %B 15th International Euro-Par Conference %2 Delft %2 Pays-Bas %8 2009-08-25 %E Springer Verlag %2 Lecture Notes in Computer Science <?xml version="1.0" encoding="UTF-8"?> <listBibl xmlns="http://www.tei-c.org/ns/1.0" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML"> <listBibl type="article"><head>Communications avec acte</head> <biblStruct type="article" xml:id="inria-00379168"><analytic> <author><persName><forename>Brice</forename><surname>Goglin</surname></persName> <affiliation> <orgName type="laboratory">LaBRI</orgName> <orgName type="team">Laboratoire Bordelais de Recherche en Informatique</orgName> <country>FR</country> <orgName type="institution">CNRS : UMR5800</orgName> <orgName type="institution">Université Sciences et Technologies - Bordeaux I</orgName> <orgName type="institution">Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux</orgName> <orgName type="institution">Université Victor Segalen - Bordeaux II</orgName> </affiliation> <affiliation> <orgName type="laboratory">INRIA Bordeaux - Sud-Ouest</orgName> <orgName type="team">RUNTIME</orgName> <country>FR</country> <orgName type="institution">INRIA</orgName> <orgName type="institution">CNRS : UMR5800</orgName> <orgName type="institution">Université Sciences et Technologies - Bordeaux I</orgName> <orgName type="institution">Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux</orgName> </affiliation> </author> <title type="main" level="a">NIC-assisted Cache-Efficient Receive Stack for Message Passing over Ethernet</title></analytic><monogr><meeting>15th International Euro-Par Conference</meeting><imprint><pubPlace><settlement>Delft</settlement><country>Pays-Bas</country></pubPlace><biblScope type="publicationDate">2009</biblScope></imprint></monogr><idno type="url">http://hal.inria.fr/inria-00379168/en/</idno></biblStruct> </listBibl> </listBibl>

NIC-assisted Cache-Efficient Receive Stack for Message Passing over Ethernet Brice Goglin () 1, 2 (2009)

Accès au texte intégral PDF

15th International Euro-Par Conference (2009)

Résumé : High-speed networking in clusters usually relies on advanced hardware features in the NICs, such as zero-copy. Open-MX is a high-performance message passing stack designed for regular Ethernet hardware without such capabilities. We present the addition of multiqueue support in the Open-MX receive stack so that all incoming packets for the same process are treated on the same core. We then introduce the idea of binding the target end process near its dedicated receive queue. This model leads to a more cache-efficient receive stack for Open-MX. It also proves that very simple and stateless hardware features may have a significant impact on message passing performance over Ethernet. The implementation of this model in a firmware reveals that it may not be as efficient as some manually tuned micro-benchmarks. But our multiqueue receive stack generally performs better than the original single queue stack, especially on large communication patterns where multiple processes are involved and manual binding is difficult.

1 :	Laboratoire Bordelais de Recherche en Informatique (LaBRI)
	CNRS : UMR5800 – Université Sciences et Technologies - Bordeaux I – Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux – Université Victor Segalen - Bordeaux II
2 :	RUNTIME (INRIA Bordeaux - Sud-Ouest)
	INRIA – CNRS : UMR5800 – Université Sciences et Technologies - Bordeaux I – Ecole Nationale Supérieure d'Electronique, Informatique et Radiocommunications de Bordeaux

Domaine

:

Informatique/Système d'exploitation

inria-00379168, version 1

http://hal.inria.fr/inria-00379168/fr/

oai:hal.inria.fr:inria-00379168_v1

Contributeur : Brice Goglin <>

Soumis le : Lundi 27 Avril 2009, 22:55:51

Dernière modification le : Mardi 28 Avril 2009, 08:21:03