Google and IBM have formed an initiative to help teach computer science students how to develop highly parallel computing applications.

Parallel computing is a way for computers to quickly carry out large tasks by simultaneously handling several different instructions through multiple processors. Researchers at the University of Maryland, for instance, developed a parallel processing desktop this summer that they say runs 100 times faster than today's PCs.

IBM and Google said that several common Internet applications, including search engines, social-networking sites and mobile commerce, often need to have tasks broken into hundreds of smaller parts that run simultaneously across different servers. They also said that parallel programming can be used for scientific purposes, such as gene sequencing and climate modelling.

To get their project up and running, the two companies have dedicated a cluster of several hundred processors that can be accessed by students over the Internet to test parallel computing projects.

The companies said that the servers will allow students to develop programs for clusters that run Hadoop, an open source application designed specifically for writing and running large applications.

Google and IBM said that the University of Washington is the first to join the initiative and that Carnegie Mellon University, MIT, Stanford, UC Berkeley and the University of Maryland will also pilot the programme.