How much input is required?

Luke

Luke

Luke

Soldato
Joined
19 Jul 2005
Posts
7,069
Location
S. Yorkshire
I see a lot of posts suggesting that output suffers when a folder has been on holiday or that work was busy etc.
Is this because the rigs are off or is ongoing attention required?

My computers are set up once and left to themselves. Does this mean I'm doing it wrong?
 
I switch mine "not on" when I'm on holiday and there is the odd crash from time to time and occasionally a hung unit
 
Last edited:
I left mine for 4.5 months once without a hiccup. Some people have their systems heavily overclocked and a WU comes along that stresses their system a bit more than existing ones causing a crash or WU failures.

For example my Q9300 is clocked at 3.1Ghz when I know it can finish WUs at 3.4Ghz I take a similar approach with my gfx cards. It's fun to see how much you can get out of your setup but obviously there are drawbacks to pushing the boundaries.
 
Yep its swings and roundabouts, remember that running S@H or Folding@home or any of the other distributed computing projects is like running Prime95 24/7

Now some people advocate that for a stable system you just need to run Prime95 or Lynx for 8 hours..

I've seen those burn in apps run for 7:58 and then crash in the last two minutes.

See what I mean?

So what I tend to do is find the potentially stable limit and then clock it back a notch or two for 24/7 use.

Helps with the heat as well. :)
 
My main rig I turned off when I went on holiday, but I left my server on.

The main rig is clocked higher as it's only really on when I'm around, so I can keep an eye on it. The server has less overclock so I can leave it for long periods without worrying about it.
 
Biffa said:
Yep its swings and roundabouts, remember that running S@H or Folding@home or any of the other distributed computing projects is like running Prime95 24/7

Now some people advocate that for a stable system you just need to run Prime95 or Lynx for 8 hours..

I've seen those burn in apps run for 7:58 and then crash in the last two minutes.

See what I mean?

So what I tend to do is find the potentially stable limit and then clock it back a notch or two for 24/7 use.

Helps with the heat as well. :)
Click to expand...
Yea 8hrs on P95 won't cut it, not for DC anyway.
I've had rigs fail upto 22hrs!, even then I've had 1 pass at 24hrs & it failed F@H within 10mins! lol
OCCT is a better tester than P95, as far as I recall I haven't had a rig fail in DC after passing a 24hr OCCT run (don't bother with linpack test though,IMO that gets the CPU unrealistically hot).
Oh & yea I do a helluva lot of testing before I let a rig lose on DC, does get boring sometimes testing so much!, but worth it in the long run :).

Btw, I haven't been here in ages, how are you guys doing in Team OcUK?:)
 
Very well indeed :) got some mad folders in the shape of the OcUk staff crew of rjkoneill and Nathwraith throwing their work machines and burn in builds into the pot, which has inspired the top 20 back into life and given me a run for my money :)

In Seti and Milkyway@home we have some recent new blood in the shape of halz and ToxicTbag among others who have taken the BOINC rug down from the wall and given it a vigourous shake upsetting us loafers in the top 10 with some MAHOOSIVE output and hardware expense :eek:

Sort of a hardware arms race in the dunjiins :)

I may be somewhat responsible as my weebeastie build and rebuild and rebuild again may have inspired some of the more extreme stuff going on
sekret.gif
 
Last edited:
There are also those who's work can require resources that prohibit folding.
For instance GPU drivers don't tend to like running CAD packages and folding at the same time, even when the CAD is not graphically demanding.
I also run CFD on my GPU, so my the folding stops so I can do real work!
 
Whats CFD?

Actually I find that anything that uses the GPU including flash can crash the folding client. :(
 
Assimilator1 said:
Yea 8hrs on P95 won't cut it, not for DC anyway.
I've had rigs fail upto 22hrs!, even then I've had 1 pass at 24hrs & it failed F@H within 10mins! lol
OCCT is a better tester than P95, as far as I recall I haven't had a rig fail in DC after passing a 24hr OCCT run (don't bother with linpack test though,IMO that gets the CPU unrealistically hot).
Oh & yea I do a helluva lot of testing before I let a rig lose on DC, does get boring sometimes testing so much!, but worth it in the long run :).

Btw, I haven't been here in ages, how are you guys doing in Team OcUK?:)
Click to expand...

We would be a lot better if you chose to change your sig to seti team OCUK :p:p
 
Lol, I'm sure you would be! ;):p

Biffa said:
Very well indeed :) got some mad folders in the shape of the OcUk staff crew of rjkoneill and Nathwraith throwing their work machines and burn in builds into the pot, which has inspired the top 20 back into life and given me a run for my money :)

In Seti and Milkyway@home we have some recent new blood in the shape of halz and ToxicTbag among others who have taken the BOINC rug down from the wall and given it a vigourous shake upsetting us loafers in the top 10 with some MAHOOSIVE output and hardware expense :eek:

Sort of a hardware arms race in the dunjiins :)

I may be somewhat responsible as my weebeastie build and rebuild and rebuild again may have inspired some of the more extreme stuff going on
sekret.gif
Click to expand...
Lol, Mahoosive!:D, 1 thing I miss about being on a US dominant forum, excellent new words like that! ;)

Nice going with your team then :), that explains why you're slowing closing in on us now in F@H! (My current main project). Although your stats don't show that yet as we're a long way off from ya atm ;).
Just got my quad on it averaging ~6400 ppd last week, funny thing is I'd be in virtually the same position in your Top 20 as I am in TAs!
 
You must log in or register to reply here.
Back
Top Bottom