Any studies on how this affects the lifetime of your hardware?
Any studies on how this affects the lifetime of your hardware?
Not that I know of. At least nothing serious and credible.
I expect this to reduce the lifetime. But not by any meaningful margin if one doesn't try to run it 24/7. All my hardware (GPU or CPU-based folding) run at acceptable temperatures while folding though. It's not that this will overshoot any recommended values. In any case the CPU based clients (normal CPU and SMP) are way more conservative than the GPU. So I reckon those are safe bets, as long as you aren't overclocking too much. It's the GPU based client that gets me a bit more worried. When active, I do get acceptable temperatures (59-60c steady on a Nvidia GTX560 Ti), but at the expense of a 40% stock double-fan. I don't feel too comfortable running the GPU so intensively for more than a few hours. So I usually run it and turn it off after 4-6 hours (that's 2 to 3 work units done for a total of 4000 points, tops).
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
I'm using the SMP client, with:
So I'm using 6 out of 8 on my core i7 920, with no affinity (so the OS can do what it thinks is best). I develop on this box all day, mostly within a VM that uses 2 cores itself.Code:[settings] extra_parms=-smp 6 ... [core] nocpulock=1
I can't even tell the FAH client is running in my normal day-to-day usage.
gg
Thanks for the correction. Very nice performance levels you have there. Got me fooled into thinking you were using the gpu fah.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
If you fold using your cpu, you'll have no problems if you keep your temps down. Heat is the killer. I strongly recommend and use a free utility program like "RealTemp", to tell you what your temps are WHILE you're folding. Set your BIOS to either throttle or shut down, 10 degrees below the manufacturer's recommendation.
I don't monitor my temps all the time, but you want to establish what your normal temp range is, and then if you hear the fans working extra hard, you have something to check it with.
Some BIOS's will show the temps, but those are idle temps - not what you want BUT perfect for validating what RealTemp is showing for temps, when the system is at idle.
GPU's are another matter - lots of them have failed while folding. They appear to be inadequately designed to handle the higher temps, for extended periods of time. Lots of people do fold with them, but I've got two GTX-280's that bit the dust from it, and I'm moderately conservative with temps - for an overclocker. Use with care, if at all.
Except for Summer's warmest days, I usually fold 24/7 for Folding@Home, but sometimes "crunch" for SETI, etc.
The Cboarders folding team has moved up 15 places in the last week, which is great considering we have just four active folders, atm.
We're about to be passed by several teams, but that will slow down a bit, as soon as my double quad rig starts returning it's work.
Congrats to Codeplug, (our fearless average points leader), Mario F, and ProgBman, for their contributions this week.
We do have signs of life.
If you would like to join in, post up and we'll walk you through the set up, and what Folding@Home, is all about.
Great team and individual folding stats (like this one), are available 24/7 from:
http://folding.extremeoverclocking.c...php?s=&t=43476
Great stats site, with updates from Stanford University (HQ for Folding@Home), every 3 hours.
Last edited by Adak; 04-10-2011 at 02:37 AM.
I've stopped for the past couple of days (and will also today) as I'm playing Titan Quest with a group of friends. LAN Party long weekend here at home. Will resume this Monday.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
So while you're burning through the WU's, how many Watts are you burning up?
A Note On Plug-in Power Meters for the UK: Review - Earth Notes
It seems like a good idea for CPU cycles which would otherwise be gone.
But it seems like an expensive hobby just to overclock a rig and run it 24/7 just so you can look good on some stats pages.
If you dance barefoot on the broken glass of undefined behaviour, you've got to expect the occasional cut.
If at first you don't succeed, try writing your phone number on the exam paper.
I'd agree that running it 24/7 would be nuts for any common home user. Including myself.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
You're uninformed, Salem.
Although molecular shape and composition is analyzed in labs, it's *very* time consuming and expensive to do it that way. Today, a huge amount of that analysis is done using sophisticated computer programs. Only some of those results are double checked by "real" analysis). Problem is, the molecules that we need to study (like the one's that let virus's into our cells, typically), are very big. The program would bring a PC to it's knee's, even today.
So distributed computing was born. Today, d/c computing power dwarfs the fastest supercomputer.
Before a drug company risks millions of dollars on a new candidate compound, it's extensively studied via computer simulation, to ensure it has a good chance of being effective, and safe. When the DOD in the US wanted likely candidates for a better Anthrax vaccine, they came right to distributed computing to find them. Sure, some projects are fun or silly even, but there are plenty of serious one's:
*finding good candidates for a Dengue Fever/Hepatitis C/West Nile and other closely related viral diseases in this class, including Hemorrhagic Fever.
*studying Muscular Dystrophy, and Huntington's Disease
*studying Childhood Cancers
*analyzing more effective way to stop malaria from spreading, given the current
situation, and assets to work with.
*climate prediction
*clean water project study
and of course, protein folding, with Folding@Home, Rosetta, etc.
You sound like a World Community Grid prospect, Salem: Roll your mouse over "research" on this page - you will be impressed. It's not just "pure" research or getting your name high up on the stats list, somewhere.
World Community Grid - Home
If a little competition with decidedly meaningless points, help people to participate more fully, and enjoy the experience, then I'm all for the competition. After all, the large majority of the folders, are not scientists, and don't understand all the details of the research they're contributing to. Giving them some way to have fun, while contributing as well, seems like a great idea to me.
Since Folding@Home started, many other molecular simulation d/c groups have been created, using a similar approach. It appears everyone wants safer, more effective drugs, cheaper, and developed faster.
In my part of the Southwestern US, we have had West Nile Virus carried in by the migrating birds, and infect, sometimes kill, our neighbors. Dengue fever is also, headed our way, through Mexico. For that reason, I'm moving most of my computer power, over to those projects, later this next week.
I like Folding at Home, but it's research is a bit too "pure" for me, at this time. Naturally, I applaud those who still work in any worthwhile project, however.
Here's an example from a post on Word Community Grid:
Now AVI Pharma in Corvallis Oregon, is also being given a large grant by the UK group "Action Duchenne", to move forward with the further refinement, and trials of some promising drugs to slow or stop DMD.
That's one of the Word Community Grid big projects: evaluate and identify the best drug candidates, for treating DMD. I'm not wasting watts, I'm heating my home with computers (well, partly anyway). In the Summer, I have to slow down or stop the computers, depending on the weather.
Duchenne Muscular Dystrophy is the most common and severe childhood form of muscular dystrophy (MD), affecting one in 3,500 boys. The disease progressively weakens muscles cells and tissues until muscle degradation is so severe that the patient dies, most often in their late teens or twenties. Scientists at Brown University in Providence, Rhode Island and the University of Pennsylvania, hope their research into the human protein, biglycan, will ultimately improve the condition of muscular dystrophy sufferers. Their studies have shown that biglycan significantly slows muscle damage and improves function in mice with the Duchenne genetic mutation. Human clinical trials will be the next step.
Human protein may help muscular dystrophy patients
Good to know that maybe one of these days families won't have to watch their child slowly waste away and die like I did a year ago. Thats why I've switched all my processing time to this project!
Ermm... I'm not moved by such goals. This is just something I do until I get bored of it. I won't deny there's some strange satisfaction in seeing my name climb up so ladder, even when that represents absolutely nothing to anyone, including myself. But mostly, I understand the project goals, am somewhat sensible to them, but don't let them dictate my actions. I do it for fun (weird that this would be fun).
The whole, you'll help save countless lifes just doesn't work with me. I won't be helping save one life even, by sitting behind a computer while it calculates protein folding. There are many more lives in danger who require more immediate intervention every day, if I truly cared. Don't take me wrong, but the canned "distributed computing saves lifes" speech we see a bit everywhere is a tad on the hypocrite side.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
Well saying what it really does--generate data for research purposes--doesn't get people to run the program. I think the competitive aspect is the whole of it anyway, so don't go harping on yourselves about wanting to watch changing statistics. It's a turf war and how good does that feel? Awesome.
Back in the really early days of distributed computing, they used to pay the people who donated their computer time to a project. Then SETI came along, and their genius was to award *points*! Although totally meaningless outside the project, the donors got totally hooked on them - and history was made.
For a successful project, one thing you MUST have, is a good stats page to show the points. The donors are crazy about the points!
The Cboarders have had a fine folding week with some milestones being recently reached:
Well done!Code:* ProgBman ----------- 250,000 * Codeplug and myself - 80,000 * MarioF -------------- 50,000
As a team, we've passed 29 other teams in the last week. These teams are not top-tier by any means, but with only four active members, that's a real accomplishment!
A great folding stats site: The Cboarders - Team Summary - EXTREME Overclocking Folding @ Home Stats
This site is updated every 3 hours, directly from Stanford (HQ for Folding@Home), beginning at midnight (Central Daylight Savings Time), 24/7.
Last edited by Adak; 04-12-2011 at 01:48 PM.