HAL - INRIA :: [inria-00312039, version 1] Current State of Java for HPC

inria-00312039, version 1

%% inria-00312039, version 1
%% http://hal.inria.fr/inria-00312039/en/
@techreport{AMEDRO:2008:INRIA-00312039:1,
	title = { {C}urrent {S}tate of {J}ava for {HPC}},
	author = {{A}medro, {B}rian and {B}odnartchouk, {V}ladimir and {C}aromel, {D}enis and {D}elbe, {C}hristian and {H}uet, {F}abrice and {T}aboada, {G}uillermo},
	abstract = {{A}bout ten years after the {J}ava {G}rande effort, this paper aims at providing a snapshot of the current status of {J}ava for {H}igh {P}erformance {C}omputing. {M}ulti-core chips are becoming mainstream, offering many ways for a {J}ava {V}irtual {M}achine ({JVM}) to take advantage of such systems for critical tasks such as {J}ust-{I}n-{T}ime compilation or {G}arbage {C}ollection. {W}e first perform some micro benchmarks for various {JVM}s, showing the overall good performance for basic arithmetic operations. {T}hen we study a {J}ava implementation of the {N}as {P}arallel {B}enchmarks, using the {P}ro{A}ctive middleware for distribution. {C}omparing this implementation with a {F}ortran/{MPI} one, we show that they have similar performance on computation intensive benchmarks, but still have scalability issues when performing intensive communications. {U}sing experiments on clusters and multi-core machines, we show that the performance varies greatly, depending on the {J}ava {V}irtual {M}achine used (version and vendor) and the kind of computation performed.},
	keywords = {{B}enchmark; {J}ava; {F}ortran; {HPC}; {P}ro{A}ctive; {MPI}},
	language = {{E}nglish},
	affiliation = {{OASIS} - {INRIA} {S}ophia {A}ntipolis / {L}aboratoire {I}3{S} - {INRIA} - {U}niversit{\'e} de {N}ice {S}ophia-{A}ntipolis - {CNRS} : {UMR}6070 - {A}ctive{E}on - {A}ctive{E}on - {U}niversity of {A} {C}oruna - {C}omputer {A}rchitecture {G}roup - {U}niversity of {A} {C}oruna },
	pages = {26 },
	type = {{T}echnical {R}eport},
	institution = {INRIA},
	number = {{RT}-0353},
    year = {2008},
    URL = {http://hal.inria.fr/inria-00312039/en/},
    URL = {http://hal.inria.fr/inria-00312039/PDF/RT-0353.pdf},
}

<?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="report"><head>Research reports</head>
<biblStruct type="report" xml:id="inria-00312039"><monogr>
<author><persName><forename>Brian</forename><surname>Amedro</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Sophia Antipolis / Laboratoire I3S</orgName>
<orgName type="team">OASIS</orgName>
<country>FR</country>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Université de Nice Sophia-Antipolis</orgName>
<orgName type="institution">CNRS : UMR6070</orgName>
</affiliation>
</author>
<author><persName><forename>Vladimir</forename><surname>Bodnartchouk</surname></persName>
<affiliation>
<orgName type="team">ActiveEon</orgName>
<country>FR</country>
<orgName type="institution">ActiveEon</orgName>
</affiliation>
</author>
<author><persName><forename>Denis</forename><surname>Caromel</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Sophia Antipolis / Laboratoire I3S</orgName>
<orgName type="team">OASIS</orgName>
<country>FR</country>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Université de Nice Sophia-Antipolis</orgName>
<orgName type="institution">CNRS : UMR6070</orgName>
</affiliation>
</author>
<author><persName><forename>Christian</forename><surname>Delbe</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Sophia Antipolis / Laboratoire I3S</orgName>
<orgName type="team">OASIS</orgName>
<country>FR</country>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Université de Nice Sophia-Antipolis</orgName>
<orgName type="institution">CNRS : UMR6070</orgName>
</affiliation>
</author>
<author><persName><forename>Fabrice</forename><surname>Huet</surname></persName>
<affiliation>
<orgName type="laboratory">INRIA Sophia Antipolis / Laboratoire I3S</orgName>
<orgName type="team">OASIS</orgName>
<country>FR</country>
<orgName type="institution">INRIA</orgName>
<orgName type="institution">Université de Nice Sophia-Antipolis</orgName>
<orgName type="institution">CNRS : UMR6070</orgName>
</affiliation>
</author>
<author><persName><forename>Guillermo</forename><surname>Taboada</surname></persName>
<affiliation>
<orgName type="team">University of A Coruna - Computer Architecture Group</orgName>
<country>ES</country>
<orgName type="institution">University of A Coruna</orgName>
</affiliation>
</author>
<title type="main" level="m">Current State of Java for HPC</title><title type="main" level="m">Current State of Java for HPC</title><imprint><biblScope type="publicationDate">2008</biblScope><biblScope type="pp">26</biblScope></imprint></monogr><idno type="HALid">http://hal.inria.fr/inria-00312039/en/</idno><idno type="HALFile">http://hal.inria.fr/inria-00312039/PDF/RT-0353.pdf</idno></biblStruct>
</listBibl>
</listBibl>

Current State of Java for HPC

Brian Amedro (

) ¹, Vladimir Bodnartchouk (

) ², Denis Caromel ¹, Christian Delbe ¹, Fabrice Huet (

) ¹, Guillermo L. Taboada (

)^a^, 3

(2008)

Access to full text
PDF

Abstract: About ten years after the Java Grande effort, this paper aims at providing a snapshot of the current status of Java for High Performance Computing. Multi-core chips are becoming mainstream, offering many ways for a Java Virtual Machine (JVM) to take advantage of such systems for critical tasks such as Just-In-Time compilation or Garbage Collection. We first perform some micro benchmarks for various JVMs, showing the overall good performance for basic arithmetic operations. Then we study a Java implementation of the Nas Parallel Benchmarks, using the ProActive middleware for distribution. Comparing this implementation with a Fortran/MPI one, we show that they have similar performance on computation intensive benchmarks, but still have scalability issues when performing intensive communications. Using experiments on clusters and multi-core machines, we show that the performance varies greatly, depending on the Java Virtual Machine used (version and vendor) and the kind of computation performed.

a –	Faculty of Informatics, University of A Coruna, Spain
1:	OASIS (INRIA Sophia Antipolis / Laboratoire I3S)
	INRIA – Université de Nice Sophia-Antipolis – CNRS : UMR6070
2:	ActiveEon
	ActiveEon
3:	University of A Coruna - Computer Architecture Group
	University of A Coruna

Domain

Computer Science/Distributed, Parallel, and Cluster Computing

Keywords : Benchmark – Java – Fortran – HPC – ProActive – MPI

Internal note : RT-0353


inria-00312039, version 1
http://hal.inria.fr/inria-00312039/en/
oai:hal.inria.fr:inria-00312039_v1
From: Brian Amedro <>
Submitted on: Monday, 25 August 2008 10:39:07
Updated on: Thursday, 7 January 2010 11:43:52