Double length WUs?
11 Dec 2006 23:47:23 UTC
Topic 192181
(moderation:
Since shortly after a mobo replacement, my higher end PC's been taking twice as long to crunch a WU as before. MY first thought was some sort of hardware issue, but I've benchmarked my CPU using sandra and the numbers are right where they should be. So I'm wondering if annother batch of extra large WUs with injected test signals have been released like was done about a month before s4 completed and I was just unlucky enough to get them at the wrong time.
Double length WUs?
)
Well, I haven't come across anything like that. Of course, it's a possibilty, but I guess in that case the project staff would have informed us. Are you sure you aren't yet another victim of themal throttling or sth like that?
RE: Are you sure you aren't
)
Actually based on farther experimenting I am. My CPU clock/power settings were identical to the old board's. This one (which is probably a bit more power hungry to begin with) expressed it's unhappyness with the load I was putting on it by not booting until I depowered my oldest HD. My old one apparently wasn't providing the CPU as much power as it claimed to be since it was running at the expected performance level, while being ~20C cooler than the spikes I was getting on my temp graph before I backed down the power.
Hmmm, that sounds more like
)
Hmmm, that sounds more like the PSU was getting unhappy. The MB/CPU just said, "Whoa, I'm not liking what I'm seeing coming in for volties."
Nice to know new gear is smart enough to realize you can't be so demanding you cause something else to fail (and new MB/CPU's can be very demanding). ;-)
Alinator
RE: Hmmm, that sounds more
)
Well, you absolutely can't have:
- too much RAM
- too highly rated power supply
- too big a hard disk
- too big a cooling fan
- too big a case
the concepts just don't make sense! :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Strange and unexplainable
)
Strange and unexplainable double-length units really exist, look at this host here
http://einsteinathome.org/host/735729/tasks
RE: Strange and
)
I don't think so. One of your quorum partners usually has short units, about ten times shorter, but also gets ten times less credit than for his two long ones. For a double-length WU, I would expect 20 times longer crunching time with 10 times less credit. To me it seems there was a problem with your computer or software.
Regards,
Bert
Somnio ergo sum
My guess is it's some edgy
)
My guess is it's some edgy problem in the science app, but really, can't prove anything ...
No, this looks like the
)
No, this looks like the science app got blocked from writing when trying to checkpoint and had to start over from scratch on the next task switch (check the RID data). The reason the time seems to "double" is the Total CPU run time for the result so far is stored in the client_state file, not the checkpoint file.
Alinator
RE: RE: Hmmm, that sounds
)
LOL, I agree except for the PSU. Even with modern high efficiency ones you should try to size the unit fairly close to your current and expected future needs. The reason is if you go too small the risk you run is obvious, overloading them is not a good idea. However, even if you run it within but near it's limits you run the risk of having it break regulation on transient conditions. When you go with too large a supply it tends to fall out of it's efficiency "sweet spot".
Alinator
Well ^^ mine is a whole lot
)
Well ^^ mine is a whole lot too big... but that is partly because I alreasy built this box with future upgrades in mind, especially a larger (hopefully Quad-Core) CPU.