📄 http:^^www.cs.ucsb.edu^research^rapid_sweb^rapid.html
字号:
Date: Wed, 15 Jan 1997 00:35:12 GMT
Server: Apache/1.0.2
Content-type: text/html
Content-length: 4871
Last-modified: Wed, 27 Nov 1996 22:58:50 GMT
<title> RAPID </title><!--<!WA0><img src="http://www.cs.ucsb.edu/Research/rapid_sweb/rapid.gif"><br><body background="brushed_aluminum.gif">--><H1>RAPID: Scheduling and Run-Time Support for Parallel Irregular Computations. </H1><h7><!WA1><img width=300 align=middle src="http://www.cs.ucsb.edu/Research/space-images/rainbow.gif"></h7><h4>This project focuses on the study ofscheduling algorithms for exploiting data, task andloop parallelism, and the development of run-time support on message-passing architectures.The fast scheduling algorithms we have developed provide effective utilization of computing resources for directed acyclic graphs, iterativetask graphs with and without cycles, and task graphs with data parallelism.The main applications are targeted atscientific computations such as sparse matrix factorization arising from numerical solutions to nonlinear equations, adaptive n-body simulationsusing the fast multipole method, and image processing. <p>We are developing a run-time system called <b> RAPID </b>which integrates automatic scheduling techniques and efficientcommunication schemes for irregular task computations with mixedgranularities on message-passing distributed memory machines.The system provides a set of library functions for specifying irregular data objectsand tasks that access these objects. It extracts a task dependencegraph from data access patterns, and executes tasks efficiently on adistributed memory machine.Our experimental results on Cray T3D and Meiko CS-2 indicatethat the system obtains promising performance in sparse matrix problems for which actual speedups have been hardto obtain in the literature. In particular, using the RAPID systemwe have obtainedgood performance for parallel sparse LU/Gaussian eliminationwith partial pivoting, which is an open parallelization problemin scientific computing literature. <p><b> <h3> Contact persons: </b> <!WA2><a href="http://www.cs.ucsb.edu/~cfu">Cong Fu (cfu@cs.ucsb.edu)</a>, <!WA3><a href="http://www.cs.ucsb.edu/~tyang">Prof. Tao Yang (tyang@cs.ucsb.edu) </a></h3><H3> Selected Publications </H3><ul><P> <li> C. Fu and T. Yang,<!WA4><a href="http://www.cs.ucsb.edu/~tyang/papers/ICS96.ps">Run-time Compilation for Parallel Sparse Matrix Computations, </a>in <em> Proc. of the 10th ACM International Conference on Supercomputing</em>,Philadelphia, pp. 237-244, May, 1996.<!WA5><a href="http://www.cs.ucsb.edu/~cfu/papers/ics96_talk.ps">Talk slides. </a><p> <li> C. Fu and T. Yang, <strong> Sparse LU Factorization with Partial Pivoting on Distributed Memory Machines.</strong><em>To appear in ACM/IEEE SuperComputing'96, November, 1996, Pittsburgh.</em><br><!WA6><a href="file://www.cs.ucsb.edu/Research/rapid_sweb//fs/rabbit/cfu/lup_docs/html_ver/sc96/index.htm"> HTML </a> and<!WA7><a href="file://www.cs.ucsb.edu/Research/rapid_sweb//fs/rabbit/cfu/lup_docs/html_ver/sc96/fuyang.ps"> Postscript </a><br>A long version is the technical report<!WA8><a href="http://www.cs.ucsb.edu/TRs/techreports/TRCS96-18.ps">TRCS96-18. </a><p><li> C. Fu and T. Yang,<!WA9><a href="http://www.cs.ucsb.edu/~cfu/papers/IPPS96.ps">Efficient Run-time Support for Irregular Task Computations with MixedGranularities, </a>in <em> Proc. of IPPS '96 - 10th Inter. Parallel Processing Symposium </em>, IEEE.Hawaii, pp. 823-830, April, 1996.<!WA10><a href="http://www.cs.ucsb.edu/~cfu/papers/ipps96_talk.ps">Talk slides. </a><p><LI> C. Fu, T. Yang, and A. Gerasoulis,<!WA11><a href="http://www.cs.ucsb.edu/~cfu/papers/irregular95.ps">Integrating software pipelining and graph scheduling for iterativescientific computing, </a><em> Lecture Notes in Computer Science</em>,Proc. of Irregular '95, Lyon, France, Sept. 1995, pp. 127-141.<p> <LI> T. Yang, C. Fu, A. Gerasoulis and V. Sarkar, <!WA12><a href="http://www.cs.ucsb.edu/~cfu/papers/ICPP95.ps">Mappingiterative task graphs on distributed-memory machines.</a>Proc. of 24th Inter. Conference onParallel Processing, Aug. 1995, Vol II, pp. 151-158.<P> <li> T. Yang, C. Fu,<!WA13><a href="http://www.cs.ucsb.edu/~tyang/papers/TR95-16.ps">Heuristic Algorithms for Scheduling IterativeTask Computations on Distributed Memory Machines, </a>Technical Report TRCS95-16,Dept. of Computer Science, UCSB, 1995.<P> <li> T. Yang, O. Ibarra,<!WA14><a href="http://www.cs.ucsb.edu/~tyang/papers/JPDC96_spdp.ps"> Performance Prediction in Symbolic Scheduling of PartitionedPrograms with Weight Variation </a>.To appear in Journal of Parallel and Distributed Computing.A short version appears in Proceedings of IEEE SPDP'95. <p></ul><H3> <!WA15><a href="http://www.cs.ucsb.edu/~tyang/papers/">More related publications </H3> </a><HR><!WA16><img src="http://www.cs.ucsb.edu/Research/rapid_sweb/up-arrow.gif"><!WA17><a href="http://www.cs.ucsb.edu/Research/PSL.html"> <b>Back to Parallel Systems Lab Home Page </b></a> or<!WA18><a href="http://www.cs.ucsb.edu/"> <b>Back to CS Department Home Page </b></a><HR><p><h3>You are visitor No. <!WA19><img src= "http://www.engr.wisc.edu/cgi-bin/counter?RAPID_MAIN"> since February 5, 1996.</h3>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -