Message boards : Number crunching : PS3 preliminary crunching numbers
Author | Message |
---|---|
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
3051 machines = 92 teraflops teraflop / machine will likely drop in the future as people don't crunch 24/7, but instead crunch part time... At 30 gigaflops per machine, this puts the PS3 solidly ahead of any CPU crunchers, especially on a performance per dollar spent (including electric bill). GPU protein folding is still more powerful and economical (raw performance and performance per dollar spent), but it is not as user-friendly as the PS3 folder. ALSO, the PS3 and GPU folders cannot work on 7/8ths of the datasets for FAH because they are not as complex as CPU folders... so we still need CPU folders, of course. We still need Rosetta folders too! |
B-Roy Send message Joined: 26 Sep 05 Posts: 26 Credit: 46,951 RAC: 2 |
interesting. may I ask where you got those numbers from? edit: I just saw your message on seti too. Here is the link: http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats Currently: OS Type Current TFLOPS* Active CPUs Total CPUs PLAYSTATION®3 102 3409 4338 |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
thanks for adding the link! It's exciting stuff. 100 more teraflops for protein folding/prediction! |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
WOW! These numbers are crazy! ... 251 Tera Flops from 10238 PS3s and rising. "The most powerful distributed computing project was SETI@Home at 280 teraflops. Now it is Folding@Home, which is at 492 teraflops and rising." (is this quote right about SETI?) Anyhow, this news is very exciting for science! |
The_Bad_Penguin Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
When I eventually break down and purchase a PS3, it will be because of its bang-per-buck (the hardware sells for ~$200 less than mfg'ing cost) for Distributed Computing. If it's only F@H, then it's F@H who gets the cycles 24/7. Not a gamer, and won't be purchasing any PS3 games. Sony's gonna lose money on me. WOW! These numbers are crazy! |
rochester new york Send message Joined: 2 Jul 06 Posts: 2842 Credit: 2,020,043 RAC: 0 |
When I eventually break down and purchase a PS3, it will be because of its bang-per-buck (the hardware sells for ~$200 less than mfg'ing cost) for Distributed Computing. its been on the news here everywhere else too i bet.....http://folding.stanford.edu/news.html#150K |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
Early Saturday morning update: PS3s: 465 Teraflops, 18992 machines. F@H Total: 716 Teraflops. The PS3s by themselves surpassed IBM's "Blue Gene" (#1 supercomputer in da world). ... AND F@H calculates flops very conservatively. A square root of a number can easily be counted as 10-20 flops, but F@H counts it as 1-2 flops. Watchers of the project are hoping that F@H's total reaches 1 Petaflop. It could happen by Sunday or Monday (unless there are many people who haven't bothered with the firmware update until getting off work Friday). |
Greg_BE Send message Joined: 30 May 06 Posts: 5691 Credit: 5,859,226 RAC: 0 |
those are amazing numbers! over how long a period of time do they cover? |
Michael G.R. Send message Joined: 11 Nov 05 Posts: 264 Credit: 11,247,510 RAC: 0 |
Very impressive. The bottom line, of course, is the amount of useful science that can be done. how limited are the PS3 WUs compared to the CPU ones? |
John Send message Joined: 18 Mar 07 Posts: 24 Credit: 0 RAC: 0 |
How exactly and what is "TFLOPS" supposed to measure? 20x faster seems to be the most conservative number I've seen. http://fah-web.stanford.edu/cgi-bin/main.py?qtype=osstats I've calculated it to be ~26x more than PCs which isn't too far off (702TFLOPS/28672)/(153TFLOPS/160524). That figure is probably inflated, considering most of the PS3 Folders are probably sitting there collecting dust, and the PC Folders are either ancient Celerons or busy doing something. === And that's just "TFLOPS". http://fah-web.stanford.edu/psummary.html There I see PS3 constituting 12/100+ WU. The PS3 WU's don't award that much (compared to Gromacs)... So how in the world is PS3 getting such a high TFLOPS number? |
John Send message Joined: 18 Mar 07 Posts: 24 Credit: 0 RAC: 0 |
http://fah-web.stanford.edu/psummary.html Also, on that link, 264 333 264 333 576 576 576 576 582 582 582 582 are the number of atoms per ps3 work unit? A few of the seemingly non PS3 work units.. 19399 6115 1135 14917 14836 91787 91787 87920 25541 18894 126006 77373 126006 308047 |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
How exactly and what is "TFLOPS" supposed to measure? See this QA item. It is a measure of computing power. For DC projects, it gives you an idea of how large the project is, and how much computing power it would take to do the same work in another mannar. So when you compare PS3 TFLOPS with TFLOPS of PCs, you are comparing the combined work of different numbers of machines. Rosetta Moderator: Mod.Sense |
John Send message Joined: 18 Mar 07 Posts: 24 Credit: 0 RAC: 0 |
Ok.. I think TFLOPS is baloney marketing hype term. http://fah-web.stanford.edu/psummary.html So what do you make of this link showing few PS3 units and few atoms per PS3 unit? |
John Send message Joined: 18 Mar 07 Posts: 24 Credit: 0 RAC: 0 |
Basically, are they saying that PS3 is crunching 73% worth of workloads or is it just measuring "perfomance"? |
The_Bad_Penguin Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Probably "old" news, but PS3 crowned king of folding at home ACCORDING TO SOME FIGURES on the Stanford Folding@home pages, the Sony PS3 is making mincemeat of all the other types of "computer" involved in the project. The page here shows 30,915 PS3 CPUs as responsible for 492 TFLOPS of processing. In comparison, some 162,197 Windows PCs manage some 154 TFLOPS of calculations. The site measures TFLOPS as the actual teraflops from the software cores, "not the peak values from CPU/GPU/PS3 specs" and notes that Active PS3s are defined as those which have returned WUs within two days. Perhaps there's something in the claims for that new-fangled Cell chippery after all. |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
The work units for PS3 involve a smaller number of atoms, yes, AND when you do a apples-to-apples comparison, they are more powerful than your quad-core folders. PS3 has only 256MB of accessible memory for crunching. The video memory isn't accessible for crunching (not sure why). This is why the PS3 WUs do the smaller WUs, but they get more work done on them and get them done faster. Like my original post, PS3 and GPU can only crunch about 1/8th of the various projects that the Stanford scientists are looking at, so they still need PCs (and Macs) for the other 7/8ths, because of more RAM and more flexibility in the types of problems that are solved. |
The_Bad_Penguin Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
PlayStation Users Boost Medical Research Stanford University's Folding@home distributed computing project has seen its capacity more than double in the last month thanks to the addition of idle processor cycles from hundreds of thousands of PlayStation 3 consoles. Total computing power of the system is now at around 700T Flops (floating point operations per second), with nearly 400T Flops of that coming from roughly 250,000 PlayStation 3 consoles, Sony Computer Entertainment Inc. (SCEI) said Wednesday. The addition of the PlayStation 3 consoles and the publicity surrounding it have also helped increase participation of PC owners in the project. The number of active PCs has jumped 20 percent in the last month, SCEI said. On Wednesday, SCEI said it has published an updated version of the PlayStation 3 application that improves calculation speed. SCEI said that it intends to support other distributed computing projects in a "wide variety of academic fields such as medical and social sciences." |
The_Bad_Penguin Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Xbox 360 not as useful as PS3 for Folding@home Microsoft's console not as capable as PS3 when it comes to folding proteins. Vijay Pande, creator of the Folding@home project, which harnesses the power of networked Playstation3s across the world to carry out essential research into Alzheimer's Disease, has revealed to Pro-G that the 360 is of limited help to his work. After being asked if he thought the power of the 360 could be useful, he said: "Possibly, although the cell processor in the PS3 is much more powerful for our calculations than the CPU in the Xbox 360." Giving details of how the power of the Playstation3 was useful to his research, Pande explained: "We are simulating key processes in protein folding and misfolding in Alzheimer's Disease. PS3's are performing aspects of these simulations, and doing so about 20 times faster than a typical PC." The program has seen a strong uptake by PS3 owners with more than 250,000 unique users having registered, delivering nearly 400 teraflops of computing power. Total computing power at a single moment is now recorded at 700 teraflops, more than double the capacity of the network before PlayStation 3 joined the program. There is still no word on any plans for the folding@home project to embrace the Xbox 360, but we're sure that many Xbox 360 owners would be more than willing to help out such a good cause. |
The_Bad_Penguin Send message Joined: 5 Jun 06 Posts: 2751 Credit: 4,271,025 RAC: 0 |
Xbox 360 could beat PS3 in Folding IN AN INTERVIEW with the San Jose Mercury News Peter Moore, corporate VP of Microsoft's entertainment business unit admitted that Microsoft was somewhat caught out by Sony's PS3 Folding@Home client. He said even Bill Gates had a conversation about "applying philanthropic processing power to big problems". But he souldn't resist a jab at Sony's endeavour. "I’m not quite sure yet whether we’re seeing real tangible results from the PlayStation 3 Folding@Home initiative," he suggested. Microsoft knows all the deficiencies of IBM's sluggish in-order triple-core PowerPC that is built inside its own console, and does not want to get soundly beaten by IBM's Cell. However, it is unclear whether the Vole of Redmond is aware that it has something far more powerful inside its own boxes. The Geforce 7900 inside the PS3 is no match for Xenos in the Xbox. Even the Sony Cell would probably end beaten by 48 vec4+scalar units hidden inside Xbox's 360 graphics chip. Folding@Home is Stream Computing at its finest, and six/seven/eight SPE units can flourish in the CPU. But when compared to the GPU, the Xbox 360 GPU would probably run in circles around Cell CPU. And then Microsoft's marketing machine might get interested in touting Folding@Home for the Xbox 360 console, since it would no longer be a race between a snail and a rabbit, as far as protein folding performance is concerned. The next question would then be, could Brook get set up running on a Xbox 360 GPU with all the limitations that Microsoft environment is using? |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
Thanks for the post. I think we will see some scientific crunching on the Xbox 360, it will be interesting to see which project. |
Message boards :
Number crunching :
PS3 preliminary crunching numbers
©2024 University of Washington
https://www.bakerlab.org