1.21 isn't bad at all. By efficiency, I meant completed time vs wall time. Also, I would suspect a SSD would make a difference.

CPU time vs elapsed time. Because of the writing to disk as well as other things, the time it takes to complete the task, runtime, is usually different than the amount of time, in real time (on the clock). Can be found under properties of the running WU.

Ahh, yes, I remember this but don't remember exactly where to find the info...



On the "ram cache" sounds like an excellent idea.
I haven't installed the SSD yet, been sick this week and haven't done much at all.
Moving the "time to write" to 1500 seconds has cured my intermittant lockups issue.

That rig is Definately chewing through some wu's. Keep us updated on the performance.

It's English I think, meaning Factor to Correct for the [estimated] Duration [of a task] ;>). WCG with server version 700 can throw this value away and do it's own estimate of how much work any device can do in the requested amount of time, but has not set it, because ... and that bit is not mine to share, but to say that something more advanced is looked for. Client 7.x would get told to ignore the DCF too with the <dont_use_dcf> flag [device owners have no control over]. Yours FTM receives work, but effectively has taken 1.2476 longer than originally estimated [the FPOPS that were put in the task header]. Since the tasks of WCG are quite variable [non-deterministic in nature], it's rather meaningless as the next task could be much shorter again. Rule is, DCF goes up fast [to protect against overcaching], and goes down slow.

Brilliant enough :?

--//--