http:^^www.cs.cornell.edu^info^projects^cam^sc96^main.html

来自「This data set contains WWW-pages collect」· HTML 代码 · 共 91 行

HTML
91
字号
MIME-Version: 1.0
Server: CERN/3.0
Date: Monday, 25-Nov-96 00:19:27 GMT
Content-Type: text/html
Content-Length: 4819
Last-Modified: Thursday, 19-Sep-96 16:34:27 GMT

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"><!--Converted with LaTeX2HTML 96.1 (Feb 5, 1996) by Nikos Drakos (nikos@cbl.leeds.ac.uk), CBLU, University of Leeds --><HTML><HEAD><TITLE>Low-Latency Communication on the IBM RISC System/6000 SP&nbsp;</TITLE><META NAME="description" CONTENT="Low-Latency Communication on the IBM RISC System/6000 SP&nbsp;"><META NAME="keywords" CONTENT="main"><META NAME="resource-type" CONTENT="document"><META NAME="distribution" CONTENT="global"><LINK REL=STYLESHEET HREF="main.css"></HEAD><BODY LANG="EN"> <A NAME="tex2html24" HREF="node1.html"><IMG WIDTH=37 HEIGHT=24 ALIGN=BOTTOM ALT="next" SRC="icons/next_motif.gif"></A> <IMG WIDTH=26 HEIGHT=24 ALIGN=BOTTOM ALT="up" SRC="icons/up_motif_gr.gif"> <IMG WIDTH=63 HEIGHT=24 ALIGN=BOTTOM ALT="previous" SRC="icons/previous_motif_gr.gif">   <BR><B> Next:</B> <A NAME="tex2html25" HREF="node1.html">Introduction</A><BR> <P><H1 ALIGN=CENTER>Low-Latency Communication on the IBM RISC  System/6000 SP<A NAME="tex2html1" HREF="footnode.html#56"><IMG  ALIGN=BOTTOM ALT="gif" SRC="icons/foot_motif.gif"></A>&nbsp;</H1><P ALIGN=CENTER><STRONG><EM>Chi-Chao Chang, Grzegorz Czajkowski, Chris Hawblitzel,</EM> <BR> <EM>and Thorsten von Eicken</EM> <BR> Department of Computer Science <BR> Cornell University <BR> Ithaca, NY 14853 <BR> <EM>{chichao,grzes,hawblitz,tve}@cs.cornell.edu</EM></STRONG></P><P><P ALIGN=CENTER><STRONG></STRONG></P><P><P><H3 CLASS=ABSTRACT>Abstract:</H3><P CLASS=ABSTRACT>The IBM SP is one of the most powerful commercial MPPs, yet, in  spite of its fast processors and high network bandwidth, the SP's  communication latency is inferior to older machines such as the TMC  CM-5 or Meiko CS-2. This paper investigates the use of Active  Messages (AM) communication primitives as an alternative to the  standard message passing in order to reduce communication overheads  and to offer a good building block for higher layers of software.<P>  The first part of this paper describes an implementation of Active  Messages (SP AM) which is layered directly on top of the SP's  network adapter (TB2).  With comparable bandwidth, SP AM's low  overhead yields a round-trip latency that is 40% lower than IBM  MPL's.  The second part of the paper demonstrates the power of AM as  a communication substrate by layering Split-C as well as MPI over  it. Split-C benchmarks are used to compare the SP to other MPPs and  show that low message overhead and high throughput compensate for  SP's high network latency.  The MPI implementation is based on the  freely available MPICH version and achieves performance equivalent  to IBM's MPI-F on the NAS benchmarks.</P><P><BR> <HR><UL> <LI> <A NAME="tex2html26" HREF="node1.html#SECTION00010000000000000000">Introduction</A><UL> <LI> <A NAME="tex2html27" HREF="node2.html#SECTION00011000000000000000">Active Messages background</A><LI> <A NAME="tex2html28" HREF="node3.html#SECTION00012000000000000000">SP Overview</A></UL> <LI> <A NAME="tex2html29" HREF="node4.html#SECTION00020000000000000000">Active Messages Implementation</A><UL> <LI> <A NAME="tex2html30" HREF="node5.html#SECTION00021000000000000000">Basic Send and Receive Mechanisms</A><LI> <A NAME="tex2html31" HREF="node6.html#SECTION00022000000000000000">Flow Control</A><LI> <A NAME="tex2html32" HREF="node7.html#SECTION00023000000000000000">Round-trip Latency</A><LI> <A NAME="tex2html33" HREF="node8.html#SECTION00024000000000000000">Bandwidth</A><LI> <A NAME="tex2html34" HREF="node9.html#SECTION00025000000000000000">Overheads</A><LI> <A NAME="tex2html35" HREF="node10.html#SECTION00026000000000000000">Summary and Comparison with MPL</A></UL> <LI> <A NAME="tex2html36" HREF="node11.html#SECTION00030000000000000000">Split-C Application Benchmarks</A><LI> <A NAME="tex2html37" HREF="node12.html#SECTION00040000000000000000">MPI Implementation over Active Messages</A><UL> <LI> <A NAME="tex2html38" HREF="node13.html#SECTION00041000000000000000">Basic Implementation</A><LI> <A NAME="tex2html39" HREF="node14.html#SECTION00042000000000000000">Optimizations</A><LI> <A NAME="tex2html40" HREF="node15.html#SECTION00043000000000000000">MPI Point-to-Point Performance</A><LI> <A NAME="tex2html41" HREF="node16.html#SECTION00044000000000000000">NAS benchmarks</A></UL> <LI> <A NAME="tex2html42" HREF="node17.html#SECTION00050000000000000000">Conclusions</A><LI> <A NAME="tex2html43" HREF="node18.html#SECTION00060000000000000000">Acknowledgements</A><LI> <A NAME="tex2html44" HREF="node19.html#SECTION00070000000000000000">References</A><LI> <A NAME="tex2html45" HREF="node20.html#SECTION00080000000000000000">  About this document ... </A></UL><BR> <HR><P><ADDRESS><I>Chris Hawblitzel <BR>Thu Sep 19 12:22:33 EDT 1996</I></ADDRESS></BODY></HTML>

⌨️ 快捷键说明

复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?