Flyer (PDF, English)      Flyer (PDF, German)
Download Intel's video on SEPARS (MOV, 213 MB)

Organization

This working group is part of the Gesellschaft fuer Informatik e.V. (GI), the German Computer Science Society. Within the Gesellschaft fuer Informatik, the working group is associated to "Fachgruppe Softwaretechnik (SWT)".

Participation is open to everyone, on an international level. We encourage everyone to participate. It is not required to be a member of Gesellschaft fuer Informatik.

Meetings will usually take place twice a year (one meeting in Germany, one in conjunction with an international conference).

Goals

The working group is a platform for communication and exchange between members from academia and industry.

The group advances on all levels the state-of-the art of software engineering for parallel systems (in particular, for multicore / manycore systems).

Motivation

As processor clock rates are stagnating, future performance increases will come from parallelism. Multicore processors integrate several CPUs on a single chip and are available for everyone at low cost. Industry predicts hundreds of cores on a single chip. Parallel computing left the niche of supercomputing and became ubiquitous in everyday life. This new situation has a fundamental impact on software development, research, teaching, and industry.

An increasing number of software engineers will be confronted with parallelism; robust approaches and tools are needed for the development of general-purpose parallel software. Exiting approaches in software engineering and parallel programming need to be re-evaluated, as they have often been developed for specific contexts. We need novel software engineering methods, concepts, models to systematically create efficient and correct multicore and manycore applications.

The slogan of the working group is "engineering of parallel software for everyone". One major focus is the development of methods and tools for the parallelization of general-purpose software for multicore and manycore platforms. Furthermore, the parallel software should exploit the full hardware potential.


Topics of interest