Is growth in the processing power of folding@Home exponential?
2nd November 2006: 0.2 Petaflops
16th Sep 2007: 1 petaflop (11 months later, 5x increase) Due to PS3s
May 2008: 2 petaflops (7 months later, 2x increase) Due to support of graphics cards via GPU2.
28th Sep 2008: 4 petaflops (5 months later, 2x increase) Due to support of graphics cardsvia GPU2.
27th January 2011: 23 Months have passed and we are only 0.3 native PFlops better than Feb 2009.
In short, yes until the end of 2008. After that we had no additional hardware, no major software advances and one huge recession to deal with. However I think that's about to change. Once the new version the folding@home client comes out of alpha/beta testing and into the main stream we should see another jump. This is because the latest version of the client utilizes multicore and the gpu by default with no complex setup and no need to download an advanced client. This will mean every new install will be a multicore cpu and a gpu install.
There are nearly 300,000 CPUs folding but only 24,000 GPUs. I will assume the folding@home considers a cpu to be 1 core so we will have to average this a bit as some computers will have upto 6 cores. I'll assume an average of 3 CPU cores per computer. So within two years of the new release I expect upwards of 70,000 GPUs to become actively folding. Or to put this another way, an additional 10 Petaflops. During that time I would hope the world has left the recession behind and had a doubling of hardware performance. So with a little luck we could be looking folding@home bringing in around 30 Petaflops in 2013 if they released the new software tomorrow.
The folding@home stats page is available here UPDATE: There is a new stats page: here.