Hi All, Second post in as many months, and still probably a newbie question I have an i7 2600K which was folding quite merrily. About 5 days ago it crashed and now I can only get about 2-3K PPD (it was at the 40K PPD). The current project is 6074, bigadv enabled (I know bonus has decreased), all cores used for folding. Is there a way to change project if that is what it is? Thanks in advance! Aegisx
You can 'change' project by deleting the work folder. But I'd highly recommend you leave it to run its course and see what happens next time you get a workunit. Deleting work units to up your PPD isn't allowed by Stanford.
I doubt it's the work unit, that is such a big reduction it seems unlikey. Processor speed ? Core voltage ? Motherboard ? All 8 threads @ 100% whilst folding ? Check your configuration file again, make sure you have -smp -bigadv -verbosity 9 set
Thanks for the replies! I can confirm: All 8 threads @ 100% whilst folding I am using FAH Tracker with the following options: "Use all CPU cores" "Enable -bigadv" Passkey set. Everything was ok. I was crunching a big WU (8995 points I think) and then FAH Tracker crashed. Since then I only have small units.
ok, that would make sense on the PPD. Is there a way to force bigadv units? I definately have bigadv set, in my log file: Arguments: -oneunit -forceasm -smp -bigadv
Not until you've finished the work unit you're already on. And then you should get a bigadv work unit if there are any available. Any reason you're using the oneunit flag?
no, not sure what it is or where it is set within FAH Tracker. I suppose the next question is whether I shoudl just go back to the command line and run the SMP client there rather than in FAH Tracker (presuming of course FAH Tracker is the source of my issue)
The oneunit flag lets f@h get a unit, finish it, upload it then pauses. So it won't download another work unit. Or that is as far as my understanding of it goes. And it might be worth a shot just using the command line version. Unless someone can think of a reason not to that is.
To take out oneunit in the Tracker - settings - configure - then tracker settings tab - under client settings on that page is one unit tab, then you can select any clients for one unit, take out the tick in the smp box to remove one unit.
Any i7, let alone a 2600K, would produce a five-figure ppd just crunching regular SMP workunits. 2-3K ppd is the production of a dualcore.
A 2600K @ 4.5Ghz running SMP (Windows) will yield approx. 25K PPD. Personally I don't use 3rd party applications to run my folding clients so am unable to comment on FAH Tracker. I would however try running the SMP client natively and see what happens.
Make sure the file client.cfg contains the CORRECT passkey. I mistyped mine a little while back and couldn't work out why I waspicking up std work units. Also, low ppd might be due to the ETA of the work unit being passed the deadline. When was the unit downloaded? What is your time per fold?
Thanks All. I will ensure the passkey is set correctly and probably move to the client rather than the Tracker... will update in a couple of days...
Hi All, I have just had a thought (scary). I have now started using the SMP Client instead of FAH Tracker. Passkey entered correctly. However it is still processing small units (if thats the correct phrase). Current Project is 6051. This is 500000 steps, and I used to get 10000000 step projects. Could it be either: A. I am just unlucky to be getting small units? B. Can a client be 'penalised' for bad practice? My machine had failed on units with 'UNSTABLE_HARDWARE'. Is it possible that as a result units may be handed out cautiously until the hardware proves itself again?
You need to return 80% of units successfully to get bonus points. Maybe if you don't fill these requirements then it wont issue you with -bigadvs