Your comments make no sense.

The most recent beta that one of my machines did had 19 WUs sent out starting at 05/16/2008 18:31:13 and ending at 05/16/2008 19:20:11. WUs took up to 3 hours to be returned. For the previous one, WUs were sent out from 05/16/2008 17:13:03 to 05/16/2008 18:23:52.

Clearly, a reliable machine running at 90% CPU would return results more quickly than one which takes 3 hours and, apart from that, it would make absolutely no difference anyhow unless it happened to be doing one of the last few given out, as they were given out over a period of about an hour.

Didactylos, if you believe I've misunderstood, then I suggest you re-read what I wrote. I've just reported what I observed. Over a period of some hours, multiple beta WUs were received by a machine set to run at 100% CPU and multiple rejections by a similar machine set to run at 90% CPU.

It seems to me that this is useful to mention, especially as there is nothing on the site which says "If you don't set your CPU to 100%, you won't get any beta work units".

To simplify what I said, the Scheduler determined that it needed the job back by say 19:00. The Server to Client interaction determined that your 90% machine would send it back at 19:06. That set the next reply: Cant make it for risk of missing deadline, therefore the host is not considered for the available work, at this moment in time.

I didn't blame anyone. I pointed out an observed situation.

However, if you want to get into blaming: WCG gets us to use boinc. If boinc is buggy (and the huge seemingly random changes in DCF are what I'd call buggy behaviour) and WCG doesn't bother to tell us about the bugs, or how to fix them, whose fault is that? If boinc's error message is irrelevant to the error, whose fault is that? If boinc changes values of important variables in a manner which is, to the observer, random, whose fault is that? Certainly not mine. We're here to run WUs, not to have to do hours of "research" every time boinc acts stupidly.

If all you're ever going to say is the equivalent of "go work it out for yourself from the total lack of documentation and given this tiny hint as to one variable that changes randomly for no obvious reason,", then it would be better if you just didn't say anything at all. Or, you know, you could actually say something useful, like telling us what exactly the relationship of DCF is to the situation and what we should set DCF to to avoid this happening or how to get boinc not to set it incorrectly. (... if that is even possible!)

FWIW:
1) The DCF of the machine that was successfully obtaining beta WUs is currently higher than the one that didn't.
2) These beta WUs all ran for exactly an hour, just the same way as standard rice WUs run for exactly 8 hours. Why would DCF make any difference at all?