Posted on 09/16/2007 6:43:18 AM PDT by texas booster
I like your tag line .. and so VERY true.
I added a WinXP to the lineup at the start of the month, and it's chugging along nicely. Here's something from Wikipedia about the PS3:
[My highlights above, all links live]PlayStation 3 cluster
Given the computing capabilities of the machine, there is some interest in using PS3 to build supercomputers for high-performance computing,[125] as the NCSA has already built a cluster based on the PlayStation 2.[126] Terra Soft Solutions has a version of Yellow Dog Linux for the PlayStation 3,[127] and sells PS3s with Linux pre-installed,[128] in single units, and 6 and 32 node clusters.[129] In addition, RapidMind is pushing their stream programming package for the PS3.[130]
On January 3, 2007, Dr. Frank Mueller, Associate Professor of Computer Science at NCSU, clustered 8 PS3s. Mueller commented that the 512 MB of system RAM is a limitation for this particular application, and is considering attempting to retrofit more RAM. Software includes: Fedora Core 5 Linux ppc64, MPICH2, OpenMP v2.5, GNU Compiler Collection and CellSDK 1.1.[131][132][133]
On March 15, 2007, SCE and Stanford University announced that the Folding@home project would be expanded to the PS3.[134] Along with thousands of PCs already joined over the Internet, PS3 owners are able to lend the computing power of their game systems to the study of improper protein folding and associated diseases, such as Alzheimer's, Parkinson's, Huntington's, cystic fibrosis, and several forms of cancer. The software was included as part of the 1.6 firmware update (March 22, 2007), and can be set to run manually or automatically when the PS3 is idle through the Cross Media Bar. The processed information is then sent back to project's central servers over the Internet. Processing power from PS3 users is greatly contributing to the Folding@home project, and PS3s have overtaken all other participating operating systems in teraflops contributed.[135][136] As of April 23, 2007, more than 250,000 PS3 owners have allowed the Folding@home software to be run on their systems, averaging over 400 teraflops and peaking at over 700. By comparison, the world's most powerful supercomputer, Blue Gene has a peak performance of 280.6 teraflops.[137]
The Computational Biochemistry and Biophysics Lab in Barcelona has launched a distributed computing project called PS3GRID. This project is expected to run sixteen times faster than an equivalent project on a standard PC. Like most distributed computing projects, it is designed to run only when the computer is idle.
Pity we couldn't get the rest of the PC makers to ship the code with their updates, the way they do all the other BS trash software they stick on the system. At least this is doing something valuable, and other than making me feel good costs me nothing to have contributing.
Thanks for keeping us informed, tex.
It does cost you in electricity if you have a more modern chip that goes into a low-power mode when no work is being done. The PS3 takes about 200W to crunch F@H.
Thanks for the pinger, shadow.
I see I’ve dropped a few more spots since last check. :(
antiRepublicrat, thanks again for folding for the team!
It's cooling down, the PS3 stays on much more now, even if it is 200W.
BTW, think I found out what's wrong with the Mac SMP client. It is a beta, and it expired. I wonder when the full version is due.
Let your kids try out SOCOM: Confrontation for PS3. It will let them be electronic Navy SEALs :)
If you're interested in tracking your folding machine(s) over the web, please Freepmail me or follow the links below.
Available features include:
Links
Please Freepmail me with any questions.
I’ve got an old box or two sitting around not doing anything. If I put these boxes back on the network and downloaded this program, would they be of any help if they are doing nothing but running this (and their OS)? We’re talking PII class machines.
A PII is slow but might finish a work unit in 50 days if you have decent memory. There are still about 10,000 Win 98 and about 50 Win 95 systems crunching away somewhere in the world.
Now, if these systems cost you money to get them working, leave them alone. I can justify the cost of electricity even for a slow system, but at some point the cost:benefit ratio won’t work. Work with 128 MB if possible.
You can try it out and all it costs will be a few discarded work units (don’t tell Stanford!). If the FAHlog.txt shows 1% every 12 hours then it will work fine.
By comparison, I get 1% every 20 - 25 minutes on my slower ATI card. The fastest GPU runs a tough work unit at about 1% every 6:45 minutes.
Make sure that you do not ask for large work units and turn off all screensavers etc.
There are hardware problems, especially memory, that will cause EUEs. Heat will do it also, and blowing out a system has cured some really hard-to-find problems.
Software issues start with spyware and virus problems and progress to Windows becoming unstable. Video drivers can also cause problems. Was your system updated by M$ last Tuesday? Ruh Roh.
Now, if you are running new work units the natural regression patterns will find that half of work units will fail. Its natural and will get better as out-of-bounds solvations are excluded.
When in doubt, go ahead and erase the existing work files and reload F@H. Better to lose one WU than a dozen.
The Folding programmers are working on v6 of the kernal. It may make an appearance this year.
Some of the programming results are visible in the updates to the PS3. Same systems, upgrade the PS3 kernal and work units are completed 20% faster.
Most betas are only two months long. I have upgraded my GPU a couple of times.
Of course they have called v5.04 console Beta for over a year.
There were 6 in a row. They included projects 2425,2426, & 2427 which are new. They all ended in the Quit 101-Fatal error: NaN detected: (ener[13]). After the first couple I blew out the dust but that didn’t help. I have run all the cpu and memory stress test with no problems. It must the new work units, like you said. I will delete the existing work file and see what happens.
Now if I could just use my Nvidia 8800 to fold on my other pc. I am starting to doubt that it will ever be supported.
Nan means “not a Number”, essentially infinity. Some NaNs are normal. A long string of them could indicate a failing component, often memory.
Every post I saw on a couple of forums referred to the NaN occurring when overclocking on these work units.
A long tine folder recommends using StressCPU2, since it uses GROMACS to abuse the CPU. If it can handle StressCPU2, then it can handle F@H.
I used to have stacks of nVidia cards but switched mostly to ATI for the folding. nVidia is finally working with F@H but it is still many months out.
btw, what is the power draw on a PS3?
I did run stressCPU2 on my folding machine and it ran for 24h without errors. I will try to start it up again b/c it just may be the new projects.
However, my new PC which I have overclocked will also run stressCPU2 just fine, but will not even complete old WU’s like the supervillans. The OC is stable with everything else, so I just don’t use this one for folding. One day I will take it back to stock setting just to see if it was the OC. This is the machine I have the nVidia 8800 on which I wish I could use.
About 245 watts when running the F@H with the rotating world screensaver.
About 210 watts when running the new F@H that came out earlier this month, and better optimization as well.
About 40 watts at idle.
One watt turned off, but no lifesaving research being accomplished.
I have read several threads and flame wars on the subject of NaNs. Its just like back in my programming PM days, trying to keep engineers and programmers from coming to blows across the table.
My opinion is to keep running F@H, and if you see too many NaNs just stop the service, erase and start over.
One of my systems had about 80 EUEs a few weeks ago. I rebooted, it had a few more then suddenly everything was fine.
I know that F@H is an excellent system tester, catching hardware failures early on. I also know that F@H will eventually find every buggy device driver in your system, and a few in the underlying GROMACS and AMBER code.
New proteins always have about 50% EUEs as the software completes the boundary testing the hard way - compute until illegal.
Ok, thanks. I will start it back up and let it run.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.