Once again, Stanford / Folding@Home, leading the way, and showing what CAN be done:

On Friday, Stanford launched the Pervasive Parallelism Lab (PPL). There's been lots of press describing it. The general plan for the lab is to develop a common paradigm for programming new architectures like GPU's, the Cell processor, Intel's Larabee, as well as multi-core CPUs. This is something we at FAH are very interested in, as we have had to have a unique code path for each of these (i.e. a separate code for the high performance part of the ATI GPU, NVIDIA GPU, PS3, and SMP). Having a single code path for all would be very, very exciting to keep FAH code development onto new hardware going smoothly.

<getting on soapbox>

In all our various discussions here at Rosie about gpu's, gaming consoles, cpu optimizations, etc., one thing that I believe has been missed in the big picture is those crunchers who crunch "efficiently".

By this, I mean using the newest (read fastest) hardware to crunch for projects that are optimized for them.

For example, with the PS3, I can crunch for PS3Grid, or I can crunch for Simap. PS3Grid is optimized to use the PS3's 1 ppe and 6 spe's, while Simap only uses the ppe. So, in terms of science done, science benefits more when the PS3 is used to crunch for PS3Grid than Simap.

Carry this over to the PC world. Most crunchers have more than one pc, and likely one is "newer" than the other(s).

So, from a science efficieny point of view, if project were to take advantage of the newest SSE instructions set, it would be selected to receive the benefit of any new 45nm Intel crunchers I may build/purchase.

And maybe there's a project that can take advantage of a dual core with a gpu. Then, if thats what I have, then that's who'll get it.

Finally, as almost an afterthought, is an old single core, maybe a Pentium D, or maybe an AMD K6.

And guess which project that old single core is going to crunch for? That's correct, the project which doesn't take advantage of the new and emerging technologies.

So, Rosie's defenders can talk all they want about about the micro-issue of not needing the increased speed / optimization on specific wu's, while they're missing the macro view that they may very well be left in the dust, with nothing but "outdated" crunchers working for them, as more and more projects start to optimize and gain the benefit from having the newest technological crunchers working on their behalf.

<off of soapbox>


Here's Stanford's presentation on the topic.

I don't know if it's already been mentioned, but it appears that F@H, with help from the PS3, has just now crossed the


2 petaflops


barrier. Congrats to F@H for recognizing the value of gaming consoles and gpu's.