And what is your proof it has anything to do with too much current draw through the VRMs?Quote:
Originally Posted by RazzleUltra
And what is your proof it has anything to do with too much current draw through the VRMs?Quote:
Originally Posted by RazzleUltra
I'm a sample size of one, so I'll have a hard job proving anything.Quote:
Originally Posted by FlawleZ
But...
The cooling of the card is adequate (temps checked under RivaTuner, even VRM). The current draw on most games is well below the levels where people are having problems. However, DoW II causes the current to hit those nasty levels for me (circa 80A IIRC?) and the D601 LED lights up and the driver restarts.
Furmark fails nearly instantly, even if I underclock the card, though older games work fine for hours.
Like I said, it's hardly conclusive, but other than just a generic 'bad card', I can't think what else it could be - All the crash symptoms under DoWII are exactly the same as those I get in FurMark, and that others with the same issue get in FurMark, so I'm assuming it's all the same thing.
there fine for me on my 4870, where furmark + occt crashed it with occp :\
Your problem does sound reasonable to believe that DoW II on your machine is indeed stressful enough. I do believe its the first case I've seen of an actual game doing so. Not even Crysis or Arma II can do this.Quote:
Originally Posted by RazzleUltra
...but on the other hand, that's exactly what gives me niggling doubts about it. I really can't see DoW II stressing the card more than Crysis, but the symptoms are identical.Quote:
Originally Posted by FlawleZ
Meh, might start pricing up an nVidia card soon and just wash my hands of it :D
If all you're getting is a driver crash/restart then your problem has absolutely nothing to do with what happens in OCCT (OCP). OCP doesn't crash the driver, it locks up the GFX and a reboot is always required to recover.Quote:
Originally Posted by RazzleUltra
Hmmm, well the D601 LED comes on every time it happens.Quote:
Originally Posted by largon
If I was running XP, I'm fairly confident it would be a hard reboot, but Vista has always seemed able to recover better with just a driver crash / recovery.
Try running DoW II in a VM with XP installed. Plenty of RTS games tend to get sketchy with maxed units and settings enabled. SupCom loves to crash when the units and AI are maxed out on 80x80 maps.Quote:
Originally Posted by RazzleUltra
Oh, I've seen various unrelated DoW crashes, but the significant one here is that it trips the D601 LED, which is a hardware, rather than software fault.Quote:
Originally Posted by FlawleZ
OS has no effect whatsoever on how the OCP crash manifests. Hence, your problem has nothing to do with the issue in this thread (OCP).Quote:
Originally Posted by RazzleUltra
And that's a fact.
Your problem is that either GPU or VRMs is merely overheating.
Largon,what's the general consensus about the crash(es) now?
Stable. 4870 reference card with the spike fan. 750/900, max 82.6A. 790/900 crashes after some time. Furmark stable too.
This 83A barrier is a setting on BIOS?
general consensus is that if you "play" furmark and occt...... reference 4870/90 isn't for you lol... :rolleyes:Quote:
Originally Posted by informal
I don't know about the consensus, but I'll try verify the OCP-theory by OCP modding my 4890 sometime, sooner or later. Mod itself is known now.Quote:
Originally Posted by informal
That being the case, what would you consider to be excessive temps on the GPU and VRM?Quote:
Originally Posted by largon
I've got a Zalman cooler on the GPU which runs IIRC well below 80C. Can't remember the VRM temps, but I've got ramsinks sat on them, and I've tried running with an open case and a fan pointing right at them.
After reading an article about GPU-Z I ran the OCCT GPU test again with my GTX 280s in SLI . GPU-Z indicated that my max current draw during this test was 89.5A. I've confirmed this with Everest Ultimate 5.02.1750 as well. The ~30A draw that I'm getting from RivaTuner must be from something else or it's a bug I guess.
PowerColor HD4870 512MB reference design, crash after 5 sec, but run games fine
how are you supposed to stress your card to the limit if you use a vm where the 3d hardware is mostly emulated? i don't think this would work out.Quote:
Originally Posted by FlawleZ
any solutions:(:
^Solution?
Don't run OCCT GPU test or Furmark.
Problem solved.
Great advice, although this concept seems foreign or hard to understand for some people. Use whatever games or programs you're actually running with the video card to test/heat it up. :/ I quit using Furmark with my HD4870/90 a long while ago. There's just no point when the VRMs get out of whack. I mean, how much furry donut can you stare at?Quote:
Originally Posted by largon
Apparently, for some people the answer is hours and hours on end. Must love that furry donut game!Quote:
Originally Posted by nafets
You gotta love the climbing graph, too...goes from 50 to 100! :yepp: Seriously, if the GPU doesn't crash in all of my games (after at least 4 hours) but crashes in OCCT GPU I really don't care until it crashes in my game. :shrug:Quote:
Originally Posted by Humminn55
So yeah.. I regrettably ran furmark to see how good my FC block was working... Big mistake... Got the crash. D601 light lit up. When I boot it takes forever to post and when posting the 4 lights light up except for D601 light... Anyone know what I can do? I struggle to get into windows, but now that I'm in all is good. Haven't tried furmark yet, and I won't... Games are also fine. VRMs don't go over 85C. Core is idle about 44C load is about 50C. Yeah, 4890 needs some better power management...
i can get higher clocks in furmark than in games but what really crashes my computer is windows screensavers.