Sony's official PS3 blog claims folding Petaflop 'barrier' broken.

PS3 contributions went from 450 teraflops to 800 with folding client version 1.2? WTH, that's a rather major perfrmance boost if the new version stands for the full increase!

Anyway - PS3 alone standing for 800tflops? That's huge. More than huge in fact when combned with this is apparantly generated by less than 34,000 active consoles.

Peace.
 
PS3 contributions went from 450 teraflops to 800 with folding client version 1.2? WTH, that's a rather major perfrmance boost if the new version stands for the full increase!

Anyway - PS3 alone standing for 800tflops? That's huge. More than huge in fact when combned with this is apparantly generated by less than 34,000 active consoles.

Peace.

Here is the source of the claim:

http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats

ps: The performance jump is indeed huge.
 
That makes the average PS3 performance 23.8 GFlops. That's up from, IIRC, about 17-18 GFlops on the initial release. So some of the increase in attained power at the moment is from people dedicating their PS3's to folding, presumably in response to 1.2. The number of active CPUs is higher than the average ~10% of total CPUs.
 
Looks like they plan to be focussing on it more as a result:

It's also interesting to think about what we want to do post-petaflop. One idea we have is to make the calculation even more accurate. With the methods we have in mind, this normally would slow us down quite a bit (say 10x slowdown), but on the PS3, it's likely that we may get a lot of that for free, as the memory access is the challenging part and in many ways with that now addressed, we may get more flops for free.

This really changes how we think of the economy of doing these calculations and is pushing us to more and more accurate models. One upshot for FAH is that this would drive the PS3's GFLOP rate (now typically 35 GFLOPS on a new GB WU) even higher, closer to the peak. Since the peak is something like 200 for the Cell, we still have a lot of room to possibly grow (although it's unclear how close one can get to the theoretical peak -- my guess is that 70 to 90 per PS3 may be the max for us). Nevertheless, even 70-90 would mean FAH would start getting close to 2 Petaflops!
 
by my count if they reach 70% of Cells theoretical max they'll be pushing over 3.7 petaflops just on the PS3 alone (assuming the same number of PS3s fold in the future).
 
When they say 70-90 they mean GFlops, not percentage of Cell's theoretical maximum.
 
"Only" 70-90 seems a big increase especially from the (already quite staggering) performance of the initial release pof tthe client.

I'm really happy and pleased and thrilled they're working on optimizing tge performance even further. Oh the joys of a closed fixed system..
Peace.
 
Considering how well the PS3 is doing, it makes me wonder just how much better things will be when they get all the GPU folding issues ironed out. (A friend had to stop his when Vista broke it.)

Meanwhile, I started folding on the PS3 from the beginning and just laugh it one of my friends who has his CPU do it on downtime, as I lapped him rather quickly. :p

Ah, nerdfun...
 
Last edited by a moderator:
Weak update, but I'm not going to look a gift horse in the mouth.

Stanford needs to start crediting PS3 WUs for more points. The disparity between PC SMP WUs and PS3 WUs is too large. PS3 is contributing 80% of the project's overall FLOPs and 24x7 PS3 folders can hope for 900 PPD at best. I know SMP folders getting over 4000 PPD with a single over-clocked quad-core.
 
Last edited by a moderator:
Actually, 1.3 is my favorite version just because of background music. I treat the entire thing like a musical screen saver. Heck... even the default music player should give me F@H as a visualizer option (instead/in addition to the planet visualizer they have).
 
Back
Top