[M203M4]

GPUs (which do not try to be a jack-of-all-trades like CPUs) are now being used as the core (or as co) processing (not simply visualization) units in a growing number of physics, chemistry, and engineering computational applications (in areas like material science, biomolecules, medical physics etc) which involve particular calculations the chips happen to be optimized for.

I don't know if this aspect is covered in the article (I'll read it later), but the GPU architecture has been found to be so amenable to so many academic, defense, and industrial computational endeavors that at least two GPU manufacturers (nvidia and ati) are now developing (minor revisions of their exisitng) chips specifically for the scientific computation market.

Just another example of supposedly "useless technology" eventually bearing gifts thanks to the twin engines of genuine human progress: science and the free market!

[Ernest_at_the_Beach]

See the last page,...page 10... for comment by the head of Nvidia regarding that GPU’s were taking over from CPU’s....aimed at Intel.

[TChris]

Just another example of supposedly "useless technology" eventually bearing gifts thanks to the twin engines of genuine human progress: science and the free market!

It's interesting how frequently the circle comes back around in the computer industry. CPUs used to need floating-point co-processors to do much math. Then, they were integrated into the CPU. Now, we have GPUs which are essentially super-duper floating-point co-processors all over again.

[texas booster]

The GPUs are fantastic specialty engines. Stanford University hosts the Folding@home project, the current version of Genome@home from the 90’s.

Free Republic has has a team for years and we now rank #71 out of 500,000 teams.

Please join if you can. Team 36120 would love to have you!