My BFG 6800GT dying? (Unique situation)

[Shamrock]

Ok, to start off, This has been happening since about mid December. I can start up a game, and it will run fine for about 20 minutes. Then the weird stuff starts happening. Sometimes it will blink to a black screen. I have to hit the windows button to get to the desktop, then back to the game to get anything back. I am enclosing a bunch of pictures, but first let me tell you what I have done to try and track down the culprit.

I have installed (and uninstalled) the following drivers:
83.40 WHQL
82.12
81.98 WHQL
81.89
78.92 WHQL
77.72

ALL of these drivers do the same thing! Also, on the desktop, if I am browsing the internet, and I happen upon a java website, my desktop blinks, and goes REALLY fuzzy, to the point you have to squint to see the text. I re-formatted. Even installed 2 versions of DirectX (9c, and 9c december edition) I have even bought a new monitor thinking that was it, but to no avail.
Should I be contacting BFG about a possible RMA?

First Image is Command & Conquer Generals : Zero Hour...Interface leaves, and so does my cursor.
http://members.hotdial.net/pbsams@hotdial.net/CCgZH.jpg
This is Champ Cap Challenge by iDT, the green stuff is suppose to be smoke.
http://members.hotdial.net/pbsams@hotdial.net/F1C-CC.jpg
Well, this is what I see in Need For Speed Underground 2
http://members.hotdial.net/pbsams@hotdial.net/NFSU2.jpg
The rest of these are rFactor, and this happens alot. Notice it doesnt have a ****pit.
http://members.hotdial.net/pbsams@hotdial.net/rFactor0.JPG
http://members.hotdial.net/pbsams@hotdial.net/rFactor2.JPG
http://members.hotdial.net/pbsams@hotdial.net/rFactor3.JPG
http://members.hotdial.net/pbsams@hotdial.net/rFactor4.JPG

Thanks
-Shamrock

[jAkUp]

Definetely looks like the card is failing. How are the temps?

[Shamrock]

Oh, another thing, I have NEVER OC'ed it. The 6800GT is plenty for my lowly 3200+ :p

temps are as follows:

core/ambient (degrees in C)
desktop: 63/42
ChampCar: 68/46
C&CG:ZH: 63/42
NFSU2: 73/49
rFactor: 73/49

[Lfctony]

Definitely looks like a bad card, RMA time. Maybe if they don't have any GTs left you might end up with a 7800GS. :)

[Son Goku]

Is that the CPU temp for the first number, or the GPU temp? Those numbers, all things said, do look a bit high, as I'd definitely be sh*ting bricks if I saw a CPU temp of 73 C... Yeah, yeah, some might be able to argue it can run so warm, but...

Reason I ask though, is with a case/mobo temp? of 46 and 49 C, I do wonder how hot the GPU on the gfx card is getting? My case and CPU temps are well below that, with the case temp measured in the lower 30s, but in the past when I had measured some gfx card temps and the CPU using a thermocouple in prior system, I had at least noticed that the GPU temp could run hotter then the CPU. Not sure in your case...

BTW, how long have you had the card?

Is this a recent problem on a card you've had for a year or so? (Could definitely be a card going bad there.)

You mention after 20 minutes of game play. Is it only after you've been pushing the gfx on that thing for awhile? (aka when the card could run hotter?)

Either way, you might want to get some cooling in your computer case to help bring those ambient system temps down a bit. In the interum, running case open, with a room fan or the like blowing into the thing should cool it down enough to rule out temp. If lower case temps doesn't improve this, definitely RMA the card then...

[TierMann]

I'm idling at 53/30c (70c max under load) with my BFG 6800GT OC right now. That's with a Zalman 700CU and AS5 though. With the stock cooler I got ~60c idle.

That isn't unusual for that card, but if you're saying it's idling at 63 and only hitting 73 max is that temp after alt+tabbing back to the desktop to check the temp or is it from a log while you're playing? If you are alt+tabbing back to windows it's quite probable that you're hitting 80c or higher in games. The temp drops very quickly as soon as you come out of 3d mode.

I'm just asking because I know my card doesn't like sitting around 80c. I don't get anything as major as your artifacts and weirdness, but I do get some artifacts if I sit ~80c for too long. Either way though, if you're getting artifacts that bad, it looks like it's RMA time.

Blue numers are my temps (GPU | Ambient | CPU)...
http://img514.imageshack.us/img514/9907/temp2ia.png

[Shamrock]

Is that the CPU temp for the first number, or the GPU temp? Those numbers, all things said, do look a bit high, as I'd definitely be sh*ting bricks if I saw a CPU temp of 73 C... Yeah, yeah, some might be able to argue it can run so warm, but...

That first number is the GPU core temp. The CPU temp usually hovers around 33c, 35 at full load. ambient inside the case is around 31-33 at full load (I have 8 case fans), how I got the temps, is in the NVidia control panel, where it says Temperature settings (You can also set how high the GPU can go before threshold exceeds).

BTW, how long have you had the card?

I have the receipt! it's 9/18/04

Is this a recent problem on a card you've had for a year or so? (Could definitely be a card going bad there.)
It started about mid December of 2005 (so maybe about 2 months?)

You mention after 20 minutes of game play. Is it only after you've been pushing the gfx on that thing for awhile? (aka when the card could run hotter?) For the corruptions, yes. But it can blink to black at anytime, and the blurry Java application thing is anytime I visit a java website.

Either way, you might want to get some cooling in your computer case to help bring those ambient system temps down a bit. In the interum, running case open, with a room fan or the like blowing into the thing should cool it down enough to rule out temp. If lower case temps doesn't improve this, definitely RMA the card then...

as said before, ambient case temp is around 31-33c, with 8 case fans (80mm), plus the twin fans on the gpu. One of the exhaust case fans are EXACTLY over the vid card, so it pulls the heat out.

That isn't unusual for that card, but if you're saying it's idling at 63 and only hitting 73 max is that temp after alt+tabbing back to the desktop to check the temp or is it from a log while you're playing? If you are alt+tabbing back to windows it's quite probable that you're hitting 80c or higher in games. The temp drops very quickly as soon as you come out of 3d mode.

alt+tabbing, I dont have a temperature program to monitor it

[Peoples-Agent]

I would RMA that card without hesitation, tell tell signs that it's screwed i'm afraid.

[ynnek]

That exactly what I saw when my card was overheating, which I suspect its the GPU itself.. wierd gemetry problems... Did you try cleaning it out, especially under the top plate that covers the heatsink? It comes off in 3-4 screws.. That solved my heat realted problems.. Even if you keep a clean case, the dust still collects under the top plate.

Although my card and systems runs normally hotter than those temps u posted, and is still stable.

Best way to check vid card temps while under load is to run rthdribl http://www.daionet.gr.jp/~masa/rthdribl/

Run that in a window, with your temp readings in another window.. You'd be surprised how hot you can get in 30 mins.


EDIT: I'm wondering if that exhaust fan right over your vid card is creating some weird turbulence effect? Maybe not..

[Geekusdeekus]

Damn, small world. I logged on here to explain the exact same problem with my Leadtek 6800GT. Although I also have a different problem. My whole screen goes fuzzy, then when I reboot, even the initial boot screen that tells you what card you have is fuzzy.

My comp has been with a local firm for a week and they can't diagnose a problem with it, yet I get it home tonight, load COD2 up and low and behold 15 mins later the same bloody thing happens.

I'm going to try that rthdribl program and see what happens when the temps go higher, I'll try and post a screen shot.

[CR0C0DILE]

did you take the cover off and clean the dust out?
the stock heat sink fan cloggs with dust every so often making the card overheat. Also applying new thermal paste wont hurt.
cards dont slowly die. they die all to once. if your card works when its cool then its still good.

[Shamrock]

Yes, I clean it thoroughly every 3-4 months, I even take off the little metal plate. I don't want to take off the heatsink and fan, because it might void my warranty. I did email support, and they had this to say:
he temperatures seem fine, and you have obviously tried several different drivers. In order to accurately troubleshoot this problem, I would appreciate if you could send the specifications of your power supply(watts,volts,amps,ect.) as well as the IRQ assignment of your card. This can obtained by going into Device Manager, selecting View, then Resource by Type, and then Interupt Request. Your card should be alone, on a spot in anywhere between IRQ 5 and IRQ 15. If this is not the case you should consult Microsoft Knowledge Base article # 299340 for manual reassignment of IRQs.

but the problem is, the knowledge base they sent me to you must either re-installs drivers (I did about 7 times!) or you must upgrade or re-install Windows XP!

my reply:
The power supply is:
INPUT
550W
Voltage ranges:103-132VAC or 206-264VAC RMS switch selectable
Input freguency:50-60Hz
Input current:10A for 115VAC, 5A for 230VAC
OUTPUT
520W
30amps @+5volts
normal current 24A
Short-circuit/ Over current/ Over voltage/ Over power/ Under voltage/ Electric shock free protection
**note** It has 2 fans to cool the power supply

The knowledge base you sent me to, is for forcing a HAL during an installation or upgrade of Windows XP. I have already re-installed Windows once (including the activation junk), and the only other way to force it is to install new drivers (Knowledge base says to update drivers in. I have done that several times already, as you can see from all the driver installs. My Video card is on IRQ16 (I have 22). On a side note, I did right click on the video card properties, and went to resources, and it said I have no conflicts.

[CR0C0DILE]

Youre right it might voied your warranty to take off the heatsink but it may be the problem. My card used a thermal pad originally. The thermal pad gets somewhat abused from constant heat and relaxes into a compressed state.
This makes it lose some contact pressure with the heatsink and graphics core.
Cleaning it off and replacing it with a good application of as5 will fix it.
That leaves you with a dilemna of whther to risk it and risk voiding your warranty. I did replace my thermal pad this way last year or so, it was completely shot.

[Shamrock]

BFG
The KB article the previous tech refered you to was only to be used if you
were having IRQ issues, as he stated. It's probably not an IRQ issue though;
there are two things that regularly cause the issues you're seeing. The first
is a bad card, and if we can narrow it down to that, we'll go ahead and
replace it for you.
The other possibility we need to look into is that of power and thermal
issues. Please download PC Wizard from
http://www.majorgeeks.com/download3846.html
and install it. Once you have PC Wizard open, click on the icon that looks
like a meter (bottom right icon in the group at the top left) and let us know
the values for:
CPU Temp (C):
GPU Temp (C):
12V:
5V:
3.3V:

my reply:

The values are:

CPU Temp (C): 25c
GPU Temp (C): 63c
12V: 11.19v
5V: 3.68v
3.3V: 3.25v

BFG's reply to that:
Looks like it's time to replace your power supply. Your 12V needs to be
between 11.8 and 12.2, and your 5V between 4.85 and 5.15, or close to there,
for decent performance. Having a 12 and 5 that low could (and quite possibly
has) fried your motherboard. I'd say replace the power supply, and if you get
the same behaviour, we'll go ahead and replace the card, as the PSU may have
damaged the card. If the new card you receive is getting the same issues,
than your power supply fried your motherboard, and you'll need to replace that
as well.

whaddaya think? Why has it lasted this long if it's the power supply frying my mobo?

I sent this back:

Then why has it lasted this long if my motherboard is fried? And my power supply also supplies all my hard drives, CD-ROM drives, CPU, RAM, Video, plus my Logitech steering wheel, Mouse, a small LED light, Sound card, wireless network card, and a spare USB add in card? all since Mid December? (2 months)

[Faster]

well power supplies can work fine, until something really drawns alot of current out of them.

it is worth a shot, a bad PSU could cause issues like this. However more likely, ur card GPU is dead.

[Shamrock]

I dont think it's my PSU, because my PSU has Short-circuit/ Over current/ Over voltage/ Over power/ Under voltage/ Electric shock free protection. If it were bad, then it would go out all at once, or alert me.

[Geekusdeekus]

Right I ran that program with my temps in another window but the card would only go up to 84 degrees C and up to that eveything was fine. ThenI played COD2 for about an hour and all was well. This morning I turned my comp on to play COD2 and about 15 minutes into it the screen went loopy and froze and the computer restarted.
I took the card to bits and re applied arctic silver instead of the normal paste, and cleaned any dust out, which surprisingly there was hardly any of. I then reinstalled the card and turned on the computer, 10 minutes into COD2 and the same happened again. :(

I have just been and bought a POV 6200 TC card (just to test if it is the card) and I have installed it and again ran COD2, all seems to be well so far, only time will tell. If the problem was actually the card it should have dissappeared now, and will give me a good excuse to buy a top card for once.

Oh by the way, I ran that diagnostic tool and all my voltages are fine except my 12v is actually 11.13 volts, could this be a casue?

[Shamrock]

yes, BFG tech said the +12v rail is too low, it shouldnt be below 11.8v. That is possible

[CR0C0DILE]

Right I ran that program with my temps in another window but the card would only go up to 84 degrees C and up to that eveything was fine. ThenI played COD2 for about an hour and all was well. This morning I turned my comp on to play COD2 and about 15 minutes into it the screen went loopy and froze and the computer restarted.
I took the card to bits and re applied arctic silver instead of the normal paste, and cleaned any dust out, which surprisingly there was hardly any of. I then reinstalled the card and turned on the computer, 10 minutes into COD2 and the same happened again. :(

I have just been and bought a POV 6200 TC card (just to test if it is the card) and I have installed it and again ran COD2, all seems to be well so far, only time will tell. If the problem was actually the card it should have dissappeared now, and will give me a good excuse to buy a top card for once.

Oh by the way, I ran that diagnostic tool and all my voltages are fine except my 12v is actually 11.13 volts, could this be a casue?

If your 12 v rail dips below 12 volts that is definitely a problem. I recommend you get a neo power. http://www.performance-pcs.com/catalog/index.php?main_page=product_info&cPath=76&products_id=1979
This one I paid 150$ for and its now 105$. Its got two 12 volt rails at 18 amps apiece.

[Geekusdeekus]

right well, the 6200TC has worked fine now for about 6 hours, including heavy gaming, so i think the 6800GT has well and truly died. My power supply is a good one though so I am dissapointed to see the 12V is so low, its actually a Tagan 2force 480 watt which cost me £80 and is only about 4 months old. Before ditching my card should I try another power supply? That said, the 6200 I have in now doesn't actually require the 4pin power supply that was plugged into my 6800 so maybe this lead was the culprit, what do you guys think?

[ynnek]

Right I ran that program with my temps in another window but the card would only go up to 84 degrees C and up to that eveything was fine.


How long did you run it for?

[Geekusdeekus]

I ran it for about 30 mins, was that enough?

[Son Goku]

That first number is the GPU core temp. The CPU temp usually hovers around 33c, 35 at full load. ambient inside the case is around 31-33 at full load (I have 8 case fans), how I got the temps, is in the NVidia control panel, where it says Temperature settings (You can also set how high the GPU can go before threshold exceeds).

OK, just checking, cause sometimes people pull the numbers from the BIOS, or some utility that measures system temps, and then just reports that. Your ambient and all isn't bad then...

I have the receipt! it's 9/18/04


It started about mid December of 2005 (so maybe about 2 months?)

This could be a problem, beings that we're now in 2006. 2 months is a bit late for a return to store however, which leaves one having to deal with the company. Many stores do exchanges, no questions asked, but require the exchange within 30 days.

BFG might or might not decide to put you through the ringer, albeit, the voltages you had claimed for your PSU do seem a bit low. Not that the PSU has necessarily gone bad, but rather it might just not be beefy enough to supply your current drawing needs, and keep the voltages sufficiently high enough. The drops are likely do to heavy load on the rails.

3.68 v on the 5 volt rail, I would most definitely question, and your 12 volt rail is too low also. Your 3.3 volt rail looks fine. You might want to try a bigger PSU, to see if this helps evelvate those voltages some. As someone said, 11.8 volts+ for a 12 volt rail would be more ideal, and as to your 5 volt, it's got hardly anymore voltage, then your 3.3 volt rail...

If a bigger PSU doesn't help, I'd definitely RMA that card, and it's possible it should have been RMA'd last year. Just get back to the tech then, and tell them the bigger PSU hasn't done a thing to alleviate the problem, if this proves the case...

[einstein_314]

I'd say your power supply is the problem. I'm not sure how your computer is running with 3.68V on the 5V rail. I've recently had a powersupply problem and my 5V rail was dipping below 4.4V. As soon as it went below 4.4, the computer turns its self off. How did you measure you voltages? Did you use a multimeter?

[Shamrock]

BFG said to RMA it, I got the email today!, but they would only RMA it, if I bought a PSU. They are saying the PSU has damaged the Video card, and to get an RMA. It is not repairable.

I would immediately upgrade the power supply because it would cause the same
problems with the card that we replaced for you. We will replace the card now
because the new power supply will not fix the issue. To set up a warranty
replacement RMA, and are currently in the US or Canada, please contact BFG via
phone at 1-866-BFG-FIXX (1-866-234-3499).

Crocodile? why $105 for a 480w PSU? I am gonna get an ANtec Truepower 550w for $83!
http://www.newegg.com/Product/Product.asp?Item=N82E16817103931
Antec TRUEPOWERII TPII-550 ATX12V 550W Power Supply
Dual 12v rails, nVdia SLi Certified, 2 PCIe connectors, 3 yr warranty, and best of all, it has NO bad things about it. Out of 170 reviews, it still has 5 eggs.

ordering it tonight, then I am calling BFG for an RMA.