Patterson Calls for a Manhattan Project in Parallel Computing
The weblog of the Computing Community Consortium has an opinion piece by Berkeley’s Dave Patterson, on the need for a major, government-sponsored attack on the so-called multicore challenge. You can read it here. (Note: I am acting essentially as the co-editor of the CCC blog.) Patterson is quite alarmed, making a jab at the “Dead Parallel Computer Society,” and arguing that we really don’t know how to cope with parallelism. His concern: unless we come up with ideas, this will lead to a “slowdown in portions of the US economy.”
I’m less certain about this, but it does seem to me that the most important place to turn for ideas is research in programming languages. Ultimately, it seems likely that software will have to be written intentionally to be parallel, and to have any hope of doing this right we’ll need new languages. Given our tremendous strengths in programming language research, I’m hoping that any future breakthroughs along these lines will have contributions from CMU.
Peter Lee @ August 27, 2008