Synergy Project Home Page |
|
Concluding Remarks The Synergy system uses a new systems software infrastructure interfacing new and conventional compilers, operating systems and application programs for scalable parallel processing. This infrastructure is suitable for crafting efficient parallel applications while maintaining the ease of programming. The benefits of stateless parallel processing(SPP) and the passive object-flow programming(POFP) are as follows.
* Note that every SPP program may contain multiple threads, if the application requires such. However, the interactions between SPP programs must go through the passive objects. With careful crafting and tuning, we have many (repeatable) experiments that can show that high efficiency can be realized using even the lowest Ethernet hardware [5,6,12]. We consider CTF the most important contribution since it lays a path to efficient automatic parallel programming. Systems of this size must have bugs. If you are an evaluator, please submit your bug report to shi@cis.temple.edu. Since 1995, I have been using Synergy V3.0 to teach parallel program design and analysis in graduate level courses (CIS673, Design and Analysis of Algorithms, and CIS669, Distributed and Parallel Processing). The Synergy V3.0 distribution tar file was compiled on SunOS, OSF/1, AIX, EP/IX, SGI, Ultrix and HP/UX. New revisions of the operating system(s) might break your compiling the system. I will be happy to assist you if you send in your requests via email to me. Acknowledgments Since 1987, many of my students, undergraduate and graduate, contributed to the Synergy project through numerous projects and directed study courses. Thanks to Doug Bagley, Kostas Blathras, Avi Freedman, Hashnian Rangwala for the programming in Synergy V1.0. Thanks to Isaac Rajkumar, Johnathan Wang, Luo Di for the programming in Synergy V2.0. Thank to the CIS Department computing laboratory staff members, Office of Computing Information Services staff members for installing Synergy on many campus computers. Thanks to the T.J.Watson Research Center, IBM corporation for the two year (1991-1993) joint study using R/6000 clusters to replace the IBM3090's. Thanks to the Office of Technology Transfer of Temple University for financial support of various kinds. Thanks to the Provost James England, the Dean's Office of Arts and Sciences, the Dean's Office of Research in College of Engineering at Temple University for various forms of financial support. Thanks to the Digital Equipment Corporation, the Hermes Financial Simulation Laboratory at the Operations Management Department, Wharton School, University of Pennsylvania, for the use of computing and networking resources. Finally, thanks to the National Science Foundation for the use of Supercomputing facilities at Cornell Theory Center the Pittsburgh Supercomputing Center.
We'd like to know what you think about our web site. Please leave your comments in this public guest book so we can share your thoughts with other visitors.
CommentsIt must be a great thing! I need to try on it.
Commentsone of the best web sites for this type of topic , that i have ever seen. john
Commentslooks great, does not work on LINUX Red Hat 7.0 Yuan Shi
|
Copyright @1995-2005 Temple University.
|