2653 seems to be talked about a lot on the forums. I guess it is a big SMP WU.
I am curious to know in general terms for a project like that, how long would it take without FAH? If Dr. Pande and collegues were scratching about the basement of a biology building with a small Best Buy budget for computers...
It would be an interesting little fact to know. And could be a useful factoid when talking to friends and co-workers that think I am insane.
How long would a project take.... say 2653
Moderators: Site Moderators, FAHC Science Team
-
- Posts: 18
- Joined: Mon Dec 03, 2007 6:50 pm
-
- Posts: 179
- Joined: Sun Dec 02, 2007 6:40 am
- Location: Team_XPS ..... OC, S. Calif
It all dependes on the PC you are folding with.
I can tell you with my experience.
On my QX6800 2.93Ghz:
OC'ed to 3.73, it runs 2 SMP_Linux WUs in about 18 to 19 hours
OC'ed to 3.43, it runs the same WUs in about 20-12 hours.
On my friend's 2.0Ghz, Core2, a Windows_SMP client finishes a WU in about 34-36 hours.
Peace
I can tell you with my experience.
On my QX6800 2.93Ghz:
OC'ed to 3.73, it runs 2 SMP_Linux WUs in about 18 to 19 hours
OC'ed to 3.43, it runs the same WUs in about 20-12 hours.
On my friend's 2.0Ghz, Core2, a Windows_SMP client finishes a WU in about 34-36 hours.
Peace
T.E.A.M. “Together Everyone Accomplishes Miracles!”
OC, S. California ... God Bless All
OC, S. California ... God Bless All
-
- Posts: 18
- Joined: Mon Dec 03, 2007 6:50 pm
There are a couple of answers, but to get any kind of accurate designation, the Pande Group will need to answer you.
Based on the time the project has been running, I'd guess that we've processed several hundred thousand WUs for project 2653. If they take an average of, say 1 or 2 days, that could easily approach 1000 years on a single late-model PC. I have no way of knowing if the project is 5% done or 95% done, though, so there's a lot I can't tell you.
It's certainly of the magnitude that would need a pretty good amount of time on the best supercomputers that are around but it's not clear how effective a supercomputer would be able to process it. Perhaps very well; perhaps not. (. . . depending on how Gromacs scales for massively parallel hardware)
Based on the time the project has been running, I'd guess that we've processed several hundred thousand WUs for project 2653. If they take an average of, say 1 or 2 days, that could easily approach 1000 years on a single late-model PC. I have no way of knowing if the project is 5% done or 95% done, though, so there's a lot I can't tell you.
It's certainly of the magnitude that would need a pretty good amount of time on the best supercomputers that are around but it's not clear how effective a supercomputer would be able to process it. Perhaps very well; perhaps not. (. . . depending on how Gromacs scales for massively parallel hardware)