THGC Needs You -Team 40051

Page 70 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
WOW. According to the Stats Page, Playstaion 3's have reached 600 tetraflops out of 850 total for all clients.
That's over 70% of total work being done. The really amazing part is that there are only 25,000 of them active, with something like Three Million in the wild.

On the other hand, nothing folds like a GPU. Only 720 active GPU are producing 42 TFLOPS. If we had 25,000 GPU's folding, that would be 1.5 PetraFlops (1,500 TFLOPS).

Depending on how the folding client fairs in the PS3 community, it could be the dominant platform for protien crunching for some time.
We could very well see PS3's eclipsing CPUs by a factor of 10 to 1 in less than a week.
 
The GPUs certainly have a lot of power, and one would think that if Stanford wanted more people to switch to using GPU acceleration, then they would heavily reward GPU clients. As it stands, people who donate a fraction of the GFLOPS GPUs offer using the SMP client (I get about 3 GFLOPS) are awarded more points/day. I've read a few posts where people have tried out the GPU client and given up on it because they were getting better points using CPU-only.

Consider that I get ~800 ppd for my SMP contribution of 3 GFLOPS while the average 58 GFLOPS donated by GPU clients are awarded something like 650 ppd. I'm awarded nearly 24 times better!

The counter to this point would be that SMP clients can run all types of folding simulations, while GPU clients only handle a specific set. This may be why Stanford encourages SMP more than GPU. This makes me wonder what kind of simulations the PS3 is assigned. If it is fully flexible like the SMP/standard clients, then kudos to it. The power of the PS3 is making me doubt the value of my 40-50 ppd 3-year old laptop's contribution.
 
A quote from the Petraflop Initiative (currently at .913 PFLOPS)

What types of calculations do the new clients speed?
There are primarily two types of calculations we perform, which differ by how we simulate water. The GPU and PS3 clients greatly speed "implicit solvation" simulations, in which water is handled mathematically in a continuum fashion (see the Wikipedia article on implicit solvation for more information). Our SMP client (discussed in our High Performance FAQ) will significantly speed "explicit solvent" simulations, where water atoms are handled atom by atom, in an explicit fashion, just like any other atom in our system. Currently, the GPU & PS3 only significantly speed implicit solvation and the SMP client only speeds explicit solvent, so each has its limits, but together they work to give FAH considerably more computational power than ever before.
Click to expand...
 
I see. So perhaps it's that Stanford values work with explicitly modelled water molecules more than the explicity continuum model, which would explain the points difference. All I can say about explicit modelling on a molecule scale is that I attended a symposium on direct simulation of fluid flow, and the limitation on the size of the simulation space is limited by memory. The limitation a year ago was to about a cubic milimetre for a fluid such as water. Water is actually more expensive to simulate because of the strong dipole effect, but then that effect is why water is as useful as it is.

Fortunately, we're simulating large molecules and surrounding materials rather than turbines.
 
Well I just set up my main rig for this the rest will come online later this week...should have 13 machines total folding for the team by the end of the week. :lol:
 
Well for what's it worh just joined this team. and will be leaving my main rig folding and backup pc's here at home for a couple weeks. total 7 computers that will be folding.
 
@ Doms1981 & Boduke: Awesome. Are you using the SMP client?
Click to expand...

Those are the command line bits under windows aye? Not yet - I'm lacking on time this morning but I'm gonna change it over later today - just wanted to get something up and going and functional for the time being.....that's on my windows machine - my linux box is offline until next week (upgrades) but that will be going as well and running the SMP client....all my other boxes in the house (for now) are single core/cpu boxes and that's the other 11 and the procs vary from a celeron 1000 to a P4 3.0ghz
 
Cool. What platform are you running? I've had a good experience with the 64-bit Linux client over the last couple of months, but the Windows version is brand new. I've finished only one WU so far with it, but it seems to be a little finicky for me still. Once in a while, if I leave the computer alone, the work will just stop. The program doesn't crash, but CPU utilisation goes to zero and no more frames are completed no matter how long I leave it. Killing the client and starting it again seems to jolt it back into working, but I've had to do this about three times a day. Such is life with Beta, but if anyone has found a fix to this problem let me know.
 
I'm running the 5.03 version on Windows.
Doesn't seem to stop left it up for about 14 hours now.

i'll see if the SMP 5.91 beta gives me the same problems as you've been getting.
 
Yeah, I meant the SMP Windows client is giving me this small problem. I've been running the regular client 100% stable for three months.

When you try the SMP version, post how it goes for you. I'm waiting for a couple more WU's to finish before I compare average points/day of the SMP client versus two single-threaded clients on a Core Duo laptop.
 
Will let you know on the progress of the SMP client. This thing is pretty cool. =) it's not everyday you get to help with something just by leaving a program running on your pc. :)
 
Yeah, I mentioned I was getting into FAH to a friend a couple months back. He's computer savvy, so he was at least interested as I described maximising ppd and everything. The last time I saw him he said he's starting to fold himself, since there's a history of Alzheimer's in his family. I'm not expecting a miracle tomorrow, but it's something you can feel mildly good about.
 
Yeah. well we just gotta hope that all of our efforts help these guys discover how to cure. or atleast prevent those things from happening. Like a very close uncle of mine who has cancerous tumor in the eye and is starting to affect the other.
 
I have 2 computers giving me the same problems with the windows SMP client (One is Vista Ultimate and the other is Win XP Pro with MCE Extentions). The error is:

[17:01:08] + Downloading new core: FahCore_a1.exe
[17:01:08] - Error: HTTP GET returned error code 0
[17:01:08] + Error: Could not download core
[17:01:08] + Core download error (#11), waiting before retry...

Anyone else having problems I seem to have been getting this for a few days now. I don't seem to have gotten any work yet (a few WU's assigned but never able to get the core).

Anyone have any idea how to fix? (I'm not using IE settings for either client, that would have been my first suggestion if ony else asked.)
 
I think I had this kind of problem before I had the .NET set up properly, but that wouldn't explain your issue with Vista so this may not be the solution.

I previously had no password associated with my XP user account, so running one of the setup scripts according to the FAH SMP instructions didn't work properly. I had to create a user account password and then it worked.
 
All accounts have paswords on my machines. I thought I had .Net 2.0 but I will try to update just in case. Is there anything special I should know about when I do that?

Still have no idea what to do for the Vista machine. I hope someone has an idea for that.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom