Reply With QuoteAll of mine are going to DF when Seti ends...
-Jt
Is was wondering on wich DC project I would spend my cpu cycles on after Seti ends?
- Seti ][
- Distributed Folding
- NFS .NET
- Others....
Any idea?
Thnx
Btw, my "farm":
P4 2.4 Ghz Win Xp Pro 512 Mb Ram 115 Gb Hd
P4 1.6 Ghz Win Xp Pro 512 Mb Ram 40 Gb HD
Last edited by cedricvonck; 08-23-2003 at 12:52 PM.
All of mine are going to DF when Seti ends...
-Jt
Thanx. But I really don't know...
I want to crunch WU that will take less then 1 hour (2 at most).
Not like DF wich would sometimes take 4 hours for 1 generation.
Thx anyway.
actually that's only for teh first gen which is gen 0, the rest of them should take at most 20 mins to finish
Yes I know.
But sometimes, even with the smaller proteins (56 - 96 amino acids) it sometimes takes +1 hour.
Thank you all for your input.
With the P4's you are running, for something they excel at and are on-par with AMD's you will be limited in your choices. Seventeen Or Bust would be a good choice as P4's kick-ass on this project. They are "good" on all the D2OL type projects Sengent , TSC and UD's Grid.org . One other medical related project Find A Drug seems to be P4 friendly as well.
-:Beyond:-
Where are you getting that information? This is the time graph from a run on a P4-2.4Yes I know.
But sometimes, even with the smaller proteins (56 - 96 amino acids) it sometimes takes +1 hour.
FreeDC Mercenary
Beyond: Thank you for your input.
rsbriggs: at this moment with the small protein, it flies right trough the generations.
but i was thinking, when they the structure configuration back to 96 or bigger it would probably take longer.
On the previous protein it would sometimes take about 100 minutes to complete a single generation. I'will try to upload a picture of it.
Ok- yes, it does take longer on larger proteins. We won't know how much longer, (given Howards recent tweaks to the program) until we run one in a month or so. The rise in minimum time between generations 91 and 134 on your graph was very typical of what I was seeing during the "memory leak" period. As memory was lost, it took longer and longer to complete a generation. Hopefully most of that is behind us now....
So remind me again what this absolutely-must-complete-generation-in-under-one-hour restriction is all about ???
Last edited by rsbriggs; 08-24-2003 at 11:05 AM.
FreeDC Mercenary
Why? Because I, as most of the participants in DC projects (IMHO) want to crank out as many wu's possible and the science of the project.
I am interesed in both sides of the dc projects:
- The project itself comes on the first place
- The speed at wich a Dc project is ran.
- And of course: the stats: how high can you get with a limited infrastructure.
That were my 2cents.
Ahhhhhh, an idealist!
I get your point, but, why should workunits > 1h be less important for science? IMHO this restriction is stupid (sorryOriginally posted by cedricvonck
Why? Because I, as most of the participants in DC projects (IMHO) want to crank out as many wu's possible and the science of the project.
I am interesed in both sides of the dc projects:
- The project itself comes on the first place
- The speed at wich a Dc project is ran.
- And of course: the stats: how high can you get with a limited infrastructure.
That were my 2cents.); you will get less points in the stats, and of course the whole project will crunsh at slower speed.
If you take DF for example, there is this thermometer thing, that counts the number of points generated. The current protein changes when the goal is reached. Of course bigger proteins take longer to be computed but that doesn't make them less important for science. It just takes longer to reach its goal. If this doesn't take too long, it's ok in my eyes.
What do you mean by that exactly? I guss you mean "how long will it take the project to be solved?"- The speed at wich a Dc project is ran.
Well, DF could take a while; the goal is to find an algorithm that can calculate how proteins fold; no easy task. I wouldn't guess this project finishes sucessfully in less then 5 years. Still, that's not what counts IMHO. Take RC5-72 for example: ~1500 years to completion (at current speed), and what do we reach by that? Nothing! That's a waste of ressources ...
wirthi:
This is Exactly what I am talking about.
Sorry'bout my english, I can sometimes not express myself good.
Ok thank you guys, I get your point!
I will crunch for DF,
Seti ][DF
Thanks to you all for your opinion!
Last edited by cedricvonck; 08-25-2003 at 12:36 AM.
Originally posted by cedricvonck
wirthi:
This is Exactly what I am talking about.
Sorry'bout my english, I can sometimes not express myself good.
Ok thank you guys, I get your point!
I will crunch for DF,
Seti ][
Thanks to you all for your opinion!
YAY !!!! Always happy to have another join us !!!
FreeDC Mercenary
LOL
Yes, I'll probably join you guys when Seti is finished.
Btw NFS.Net project you mentioned would be good candidate for
yer machines too - it needs lotsa RAM...and as soon as it gets
from beta, I'm going to nag admins to add it ().
But yes, DF is good choice too...
rm -Rf /
Great news! Glad to have you aboard!
You are quite right about the two projects:Originally posted by Supp
Btw NFS.Net project you mentioned would be good candidate for
yer machines too - it needs lotsa RAM...and as soon as it gets
from beta, I'm going to nag admins to add it (
DF is going to run at my 2.4 ghz.
Nfsnet is going to run at my 1.6 ghz
Posting Permissions
