Quote:
Originally Posted by MacJunky Well, if my board was in a case and I had a better PSU I would be better off than I am now ( http://img294.imageshack.us/img294/6864/00000332up9.jpg heh, random camera flash strikes again ).
Thing is, I can't finish my case modifications until I get a new PSU.
And the problem with that is the PSU money will have to come out of the "whole new computer" money. |
I think you would be much better off saving the $ for a "whole new computer" since it will likely fold a whole lot better than the g4 will.
I am certainly no mac expert, in fact not even a noob! ... but from what i have read of others folding on the g4's, any low end c2d based rig, either mac osx or windows, is going to produce a ton more since a dual core will let you run smp. There are smp clients for osx, windows and 64-bit linux too.
In the meantime, don't worry about not running the folding client on the g4 ... if most of your wu's are error'ing (eg Early Unit End errors - EUE in folding jargon) then you will will do the scientific effort a favour by not folding on it.
Every time a wu is returned as an EUE, it is reassigned to several other computers .. which means that more than 1 other computer will rerun the same wu just to prove your hardware was most likely at fault. Which wastes their time cuz they could have been crunching other wu's.
Furthermore, each wu is part of a sequence, so the next one cannot be assigned until the previous one is successfully completed. So a hardware caused eue can add many days to how long it takes to complete a project.
Some eue's are normal just because of the way the simulations are set up but these are pretty uncommon since they are usually beta tested before release. The researchers use the reassignment success/failure to determine if the eue was due to hardware instability, or the simulation itself.
Hope this helps