Validation errors
1 Jun 2008 19:51:21 UTC
Topic 193705
(moderation:
Hi,
I've been getting some validation errors on some of the results lately, and while I haven't been paying that much attention, I haven't seen it happen before. The last few days have been very hot, and I've upgraded my GPU which made my computer run quite a bit hotter as well, so I thought perhaps overheating might be the culprit.
However I usually run two cores on einstein and two cores on cosmology@home (I have a quad core), and I've had no issues with cosmology. I also tried running Prime95 an entire night and it didn't have any problems (one of the failing wu's ran during the previous night). Other than einstein@home, I haven't noticed any instability.
Could there be an issue with my computer, or is it normal to get a few validation errors here an there?
The work units in question are 39834210, 39835506 and 39873744
Validation errors
)
Thanks for the report.
Validation errors are very, very rare. Three of them on a single host within only a few days is definitely not a good sign. Checking the cooling might be a good idea.
CU
Bikeman
RE: Validation errors are
)
:( Oh well, I'll have to figure something out then.
Wonder what the einstein app does that causes it to fail though, since everything else seems to work flawlessly. I've been gaming while running prime95/stresscpu v2, and no glitches.
Thanks!
RE: RE: Validation errors
)
If you recently upgrade to a more Power hungry GPU, are you sure your PSU (Power Supply Unit) can handle it. Might be something to look at.
Unstable Vcore or any other Voltage can issue computation errors.
As the core temp can...
You are running 4.26, not the last app, but still putting a lot of pressure on the CPU temp.
I haven't ran it recently, but from what i recall, and as unusual as it seems at first look, Prime95 was putting a lot less pressure on the CPU than EAH.
And it was with an older app than 4.26.
With all the Apps optimization over the years, EAH has become one dam good testing way to put your system under as much pressure as possible.
I don't know Cosmology very well, but i'm pretty sure EAH is a lot more CPU intensive.
Did you check your CPU temp with Prime and EAH ?
Did you do the same with various Voltages your system is able to monitor ?
God created a few good looking guys.. and for the rest he put hairs on top..
Thanks for your
)
Thanks for your feedback.
I've run extensive benchmarks, gaming while using StressCPU etc, and I can't find a single fault except for these non-validated WU's. When using StressCPU, core temps are 3-5 C higher than einstein manages. PSU is a Corsair 620W, so should be ok (I got a 8800 GT).
Could it be that it's not a heat issue, but instead that I manually stopped the BOINC service? I've done that a couple of times the last few days...
I think I may have found the
)
I think I may have found the culprit. My ram's SPD says the timings should be 4-5-5-15 (DDR2, PC6400) yet for some reason my motheboard has decided to use 5-4-4-12. I'm guessing einstein stresses the ram much more than prime95 and certainly the pure cpu tests. I'm gonna try showing my BIOS who's the boss :)
In any event, thanks for the assistance :)
WIth prime95 it depends what
)
WIth prime95 it depends what tests you're using. The small CPU stressing ones don't use ram at all, the big ones probably load it harder than einstien does. To really stresstest your ram you should use memtest86. It's a bootable CD, so you just stick it in and let it do it's thing. Run at least 3 passes to be sure everything is OK. WIth a newer machine this may be an overnight task.
http://www.memtest86.com/
I had one validation error.
)
I had one validation error. too. It´s my first error since months.Nothing changed on this machine. It´s a only one month old laptop. Here is the WU.
It's not sure yet whether
)
It's not sure yet whether your result is valid or not, at the moment I write this, your result did not agree within the defined tolerances to the first submitted result. The next result will decide who get's the credit. The way BOINC displays this is quite misleading.
CU
Bikeman