I consider the bit-tech readership to be some of the brightest techies in the world, so I figured I'd ask for help here. I've been discussing my problem with some bfgtech support people, but their responses are very slow and I question their line of thinking. Instead of rewriting it all, I'll just paste my interactions with them. me: Hello, I own a Geforce 8800gtx OC2, which I got in May. At that time, I ran 3dmark06 and got a score of over 10,000. I recently ran it again and only scored 6,000. I of course tried all the obvious things up to and including completely reformatting my hard drive, so the computer is in exactly the same condition it was in in May. I am lead to believe that the problem lies in my video card because in 3dmark06's details screen, my cpu score is the same now as it was in May, but my SM 2.0, SM 3.0, and overall 3dmark score are far lower now. I have noticed that my performance in games is not what it used to be, which is what made me want to run 3dmark06 recently. So I know it is not a problem specific to the benchmark, it is some kind of over-arching issue. Is there any reason at all that a video card would run over 30% slower 6 months after buying it? The fan is blowing perfectly well and the card is not dirty at all. them: We will need to observe what the voltages are running at in real time. There are two ways to perform this. You could go into the BIOS of your motherboard and under the advanced chipset tab, there may be an option for voltage monitor. This will show you what the 1.5V, 3.3V, 5V, and the 12V rails are operating at. me: my voltages are as follows: 1.5V = 1.6 (I assume this is the processor voltage) 3.3V = 3.59 5V = 4.63 12V = 11.72 Also, it's worth mentioning that the GPU temperature reads as 63 degrees C. them: Your +5v and +12v are both low for what you want them to be at. Can you check those voltages in your BIOS? me: Yeah, my BIOS reads them as 12.09 and 4.78. After booting back into windows and rechecking PC wizard, they're currently read as 11.72 and 4.63. Would such a small power irregularity be able to cause such a big performance change? I'm using a Seasonic 700watt PSU, which I figured would be powerful enough and also a good brand name. them: Your +5v is running lower then we would like it to be at. This can cause problems with your graphical display. Can you test the card in a different system? me: I don't have another system to test the card in, but I had a better idea. I have a PSU tester that actively gives the voltages of everything. Well, the PSU has completely normal voltages when it's running outside of a computer, none are low and a couple are a tad high, but well within acceptable ranges. So, I then stuck a different PSU into my computer and checked its voltages. They were almost exactly the same as those I've already sent you, using my first PSU. Sorry if this is getting confusing. Basically, my PSU has to be OK and it seems like the problem is almost guaranteed to be in either the motherboard or the video card. I'm stuck on how to test which it is, since I lack another computer with a pci express slot in which to test the video card. I really appreciate you guys' help, and you're really confirming all the good things I read about your company before I bought my video card. I really hope we can get this fixed, I'm itching to play some Crysis on high detail =) I haven't heard back from them yet. I know that's a lot of text, but I figure it's easier than you guys possibly asking all the same questions and me giving all the same answers. Anyone have any bright ideas? I'm really stumped on this one. =/
just a quick thought, have you checked the 8800's heatsink for dust? just wondering if its getting too hot. That most likely wont cause the rating to drop from 10,000 to 6,000 but its something to check anyway. what make and model psu do you have anyway?
He mentioned the card isn't dirty and the fan is blowing well so guessing all dust has been cleaned away. Have you tried using different drivers to see if that makes any difference? Is there anyway you can observe the temperatures of the card when it's put under load? I'm thinking does it have any thermal throttling like some CPUs have, no idea if gfx cards have that on them?
My PSU is a Seasonic 700watt, which is both a good brand and a high enough wattage for my system. I also tested a second PSU, a Seasonic 500watt, in my system and it does exactly the same thing, so the PSU is good. The card is pretty clean and the fan spins without a problem. I have not tried using different drivers, I suppose that's worth a shot only because I don't know what else to do. I am able to view my gpu temps in real time using speedfan and my g15's screen. My GPU didn't even reach 40 degrees C during a full run of 3dmark06...so I think I'm good thermally. bfgtech now wants me to just RMA my card, but I would much rather figure out if anything else could be causing this, since my card appears to be in good shape.
Sounds like the card's faulty to me - 40'C is far too cool for an 8800GTX. What I'd say is to RMA it - it sounds like the BIOS has gone mad (or another fault to a similar effect) and it's stuck in low-power 3D mode... The drivers should report it, but again, when crap goes wrong, it does a tricky job.
Does it void the warranty if you flash the gfx card bios? Could be worth a go? But as Krikkit said, that temp seems suspect, just too cool. Not sure if this happens with hardware, imo it's either working or it's not... but could it be like partially damaged so it still works just not too max capacity? Kinda like losing an engine I guess, if you have two, so to speak lol.
You guys are right, I agree that the card should be running hotter than it is. I suppose I'll go ahead and RMA it. I'll be without a PC for a week or more now =( Dunno how I'm going to live. Do you guys think that maybe it's been running too hot all this time and that maybe it became damage due to excessive heat? I don't recall seeing it running too hot, but I don't check the temps very often either. My concern is that maybe I should figure out a way to get some more air flow to it so that my replacement card doesn't have the same problem in 6 more months. Currently, I have 2 intake fans up front, 2 exhaust fans in the back, and 1 exhaust fan on the top of the case. None of them really go near the GPU I guess. Think it's worth modding another fan into the case to blow across the GPU or would you say it's more likely just a defective card and a freak occurrence? Thanks for the help guys =)
Well, I RMA'd my video card and have gotten a new one in the mail...and same damn thing is happening. So I installed video card drivers from May and...same damn thing again. I've run out of ideas. I guess next step is replacing my motherboard, unless anyone else has a better idea.
As others stated, 40°C temp on air is impossibly low for that card. I get 40°C on my 8800GTS under load, but it's also on water.
Is there nothing you can remember changing that caused this? For example upgrading you chipset drivers or something? Sorry if you already mentioned it but what motherboard are you running? Does it have 2 PCI-E slots? Could try the card in the other slot? Maybe it is your power supply? Unfortunately it seems your going to have to do some hardware replacement just to find the problem. What makes me curious tho is that your card was accepted for RMA and I thought they tested them? So surely your card must have been bust? Good luck
i wouldn't test a card that got sent back to me. why would someone send back a card that is working? it takes time and money to fully test a card that someone was willing to take out, spend the money to ship it back, and go without it for a while. if you tried a different psu, and a new video card. I am thinking it must be a motherboard issue. Have you done any overclocking? if you did, did you lock the pcie(whatever that option is called. pci bus or pcie bus or something with pci in it) at 100mhz or leave it on auto?
I appreciate all the comments. Since May, when I built the system, I have done no overclocking or anything. I am using an Asus Crosshair mobo with the Nforce590 sli chipset. It does indeed have a second video card slot, I will test the card in that slot in a bit. I know my power supply is good for two reasons. One is that I swapped in a different one and got the same results. Also, I have a hardware tester from frozencpu that says my voltages are ok. As far as whether they're OK by the time they go through the mobo to the video card, I can not say. As for the pcie bus, it is set to default setting, which I'd assume is auto.
Well, I ran 3dmark06 with the card in the other pcie slot and the score was still around 6000. I think I may just buy a different motherboard, I don't know what else to try.
just thought I'd let you guys know that it ended up being the motherboard. A quick switcheroo and now I'm scoring in the 12,000's in 3dmark06. I hate problems that can only be solved by trial and error =/ Thanks anyway though.
