In the area of Computing Architectures there are some well known discontinuity-inducing trends staring us in the face. The entire computing community is planning for multi-core processors, a necessary order of magnitude(s) increase in the performance/power ratio, ‘failure is an option’ with the advent of millions of cores … and one of the holy grails, easier paralleling programming at scale. Adapting to these trends and necessities is tough and will require non-linear thinking, not just extrapolations of current trends.
Statement like this are, of course, motherhood and apple pie. Computing architecture researchers have faced all of these challenges for years and there are numerous projects forging into the future to address them, among them the Department of Energy’s Exascale efforts and the nascient program from DARPA, the Ubiquitous High Performance Computing Program.
However there is a renewed need to build a community of computing architecture reseachers across the country to give an avenue to the large number of new people in the field to bring their ideas to the national stage.
The CCC sponsors workshops (http://www.cra.org/ccc/vision.php) to help researchers put together visions for research programs. Josep Torrellas of the University of Illinois and Mark Oskin of the University of Washington recently held a workshop in San Diego on Advancing Computer Architecture Research (ACAR), February 21-23, 2010. The theme of this first of two workshops was Failure is not an Option: Popular Parallel Programming. This is a follow-on to the Computing Research Association Workshop in 2005. The workshop had representation from academia, industry and national laboratories.
Torrellas and Oskin asked of the workshop participants:
- What will computing platforms look like in 15 years?
- How will they impact the socio-human condition?
- What are the major research challenges that must be overcome to create these platforms?
And are synthesizing the discussion around two goals:
- Articulate an agenda and roadmap for computer architecture research to address the challenges above.
- Create excitement and community building for computer architecture research across academia, industry and national labs.
Workshop summaries and proceedings will soon be available at
http://www.cra.org/ccc/acar.php.
Submitted by Bill Feiereisen for the Computing Community Consortium.
