Object | Projects |
Papers | Contact
As manufacturers have encountered difficulties to further exponential increases
in clock speeds, they are increasingly utilizing the march of Moore’s law to
provide multiple cores on a single chip. The multicore processor is rapidly
becoming the mainstream on desktop-computing machines with quad-core shipping
now and 16 core system within two or three years. Tomorrows computers will have
more cores rather than exponentially faster clock speeds. The architectural
changes benefit only those software with thread level concurrency in mind and
therefore have little value for most existing
software. Consequently, the good time when people merely rely on the faster CPU
clock to speed up the software execution is over, the software must be able to
exploit the the parallelism and concurrency somehow so that it can take
advantage of multicore resource.
Service-oriented architectures(SOA) are gaining popularity due to their standards-based and highly customizable architecture. Web service technology is the mainstream to implement a service-oriented architecture. Web services itself is a set of XML based standards: WSDL as the interface description language, SOAP as the communication protocol and so on. However the performance of the web service architecture is usually a concern due to the expensive and complicated XML processing tasks. Multicore provides an opportunity to address the performance issue, but it also is a challenge to exploit the parallelism effectively. In this project we are interested in the parallel programming model, pattern and paradigm which can help improving the performance of the Web Service and XML processing by leveraging the multicore resource.
a general
parallel XML processing C# library, containing the parallel traversal/searching,
parallel serialization/C14N, parallel pull-parsing/DOM building, etc.
is the first
parallel XML parser in C++ , built upon libxml2
| Wei Lu | Last modified: $Id: index.htm,v 1.5 2008/04/18 04:40:35 welu Exp $ |
