-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
BFG 8800GTX locking up
- Thread starter Meridian
- Start date
Right, time to thrust this little problem back into the limelight...
After a few days of fiddling, this is what I have learned:
1) The clear the problem I need to power the machine of for at least fifteen seconds. I've not noticed any difference between leaving it off for thirty seconds and thirty minutes. This suggests heat is not the issue.
2) Once powered up again, the problem stays away for about two to three days (remember, the machine is always on running SETI and/or Einstein), then reappears. It does not need to be intensive 3D graphics: even Media player, or several 2D graphics windows open at once will do it. Then the display freezes (actually it seems to be freezing for about a minute, then releasing for about two seconds, then freezing again etc. It keeps that up for about five minutes, then BSODs with a NV4_DISP.DLL error).
2a) That time of two to three days is unrelated to what the computer is used for. It could be ten hours of Oblivion or two hours of surfing - the problem will not come back any sooner or any later.
3) NV4_DISP errors appear to be the software equivalent of an engine management light: "Hi. Something has gone wrong, but we won't tell you what or where!" Lots of things cause this error.
4) Once the error is there, reboots do not cure it even temporarily. However, if the problem is not there, reboots don't bring it on. Once there, only the switch off trick works - and it always works.
5) No combination of drivers, nTune, RivaTuner or lack of them makes any difference one way or the other. I've not tried the certified 162.18s though, only the beta. However, I've not tried to roll back Direct X to the previous version.
Along the route I have also discovered that neither nTune nor RivaTuner can control the fan. nTune doesn't even give you the opion. RivaTuner does, but then ignores any setting except 25% and 100%. I have managed to turn the duty cycle up though. Interestingly, there is no evidence that the fan every runs at any speed except 2400rpm - both in 2D, and running Oblivion in a window at 1024x768 (so I cans see the fan speed and GPU temp on the toolbar).
However, that said, running Oblivion doesn't cause the GPU temp to go above 52oC anyway - assuming RivaTuner is reporting the temp correctly. Mind you, nTune agrees with that temp. As I said, it doesn't look temp related.
I've also fiddled with the CPU overclock, but that looks totally stable. This fault is NOT related to ambient temperature, whereas if it was an overclocking issue you'd expect more problems at higher ambient temps. But just in case I turned the o/c down - it made no difference. Neither did turning the voltages either up or down.
The nearest way to describe what appears to be happening is: a buffer somewhere slowly fills. Once full, the system crashes. rebooting frees only a small part of that buffer, which soon fills up again. The rate at which the buffer refills is then dependant on the amount of graphics rendered - but ONLY once it has filled the first time. Once full, powering down long enough to discharge the capacitors also seems to clear the buffer. That's just an idea of the way the system behaves.
Now, time to try a suggested Windows Hotfix plus the new drivers...
M
After a few days of fiddling, this is what I have learned:
1) The clear the problem I need to power the machine of for at least fifteen seconds. I've not noticed any difference between leaving it off for thirty seconds and thirty minutes. This suggests heat is not the issue.
2) Once powered up again, the problem stays away for about two to three days (remember, the machine is always on running SETI and/or Einstein), then reappears. It does not need to be intensive 3D graphics: even Media player, or several 2D graphics windows open at once will do it. Then the display freezes (actually it seems to be freezing for about a minute, then releasing for about two seconds, then freezing again etc. It keeps that up for about five minutes, then BSODs with a NV4_DISP.DLL error).
2a) That time of two to three days is unrelated to what the computer is used for. It could be ten hours of Oblivion or two hours of surfing - the problem will not come back any sooner or any later.
3) NV4_DISP errors appear to be the software equivalent of an engine management light: "Hi. Something has gone wrong, but we won't tell you what or where!" Lots of things cause this error.
4) Once the error is there, reboots do not cure it even temporarily. However, if the problem is not there, reboots don't bring it on. Once there, only the switch off trick works - and it always works.
5) No combination of drivers, nTune, RivaTuner or lack of them makes any difference one way or the other. I've not tried the certified 162.18s though, only the beta. However, I've not tried to roll back Direct X to the previous version.
Along the route I have also discovered that neither nTune nor RivaTuner can control the fan. nTune doesn't even give you the opion. RivaTuner does, but then ignores any setting except 25% and 100%. I have managed to turn the duty cycle up though. Interestingly, there is no evidence that the fan every runs at any speed except 2400rpm - both in 2D, and running Oblivion in a window at 1024x768 (so I cans see the fan speed and GPU temp on the toolbar).
However, that said, running Oblivion doesn't cause the GPU temp to go above 52oC anyway - assuming RivaTuner is reporting the temp correctly. Mind you, nTune agrees with that temp. As I said, it doesn't look temp related.
I've also fiddled with the CPU overclock, but that looks totally stable. This fault is NOT related to ambient temperature, whereas if it was an overclocking issue you'd expect more problems at higher ambient temps. But just in case I turned the o/c down - it made no difference. Neither did turning the voltages either up or down.
The nearest way to describe what appears to be happening is: a buffer somewhere slowly fills. Once full, the system crashes. rebooting frees only a small part of that buffer, which soon fills up again. The rate at which the buffer refills is then dependant on the amount of graphics rendered - but ONLY once it has filled the first time. Once full, powering down long enough to discharge the capacitors also seems to clear the buffer. That's just an idea of the way the system behaves.
Now, time to try a suggested Windows Hotfix plus the new drivers...
M
Me back again.
I finally got fed up with this on Saturday and did a full from-scratch re-install of Windows. I loaded XP SP2, all the patches, DX9.0c August 06 build (which had worked for three months on this machine) and the 162.18 drivers. I installed Oblivion for testing but never played it. Because while watching a short film in Media Player the whole gfx locked up again.
So I've e-mailed BFG to see if I get the brush-off (or "you must be using the wrong drivers").
M
I finally got fed up with this on Saturday and did a full from-scratch re-install of Windows. I loaded XP SP2, all the patches, DX9.0c August 06 build (which had worked for three months on this machine) and the 162.18 drivers. I installed Oblivion for testing but never played it. Because while watching a short film in Media Player the whole gfx locked up again.
So I've e-mailed BFG to see if I get the brush-off (or "you must be using the wrong drivers").
M
An update, just to bore everyone...
BFG RMA'd the card saying they thought that the problem was overheating video RAM. I don't believe that it was, but it's generous of them anyway. The replacement card arrived two days after I sent them mine, so their service is good.
Guess what: either the same problem is there, or else its junior cousin. The system doesn't do a full lockup, but instead turns into a one-frame-every-twenty-seconds slide show. I can sometimes quit out of it, but not always. As before, if left off for an hour the problem stayed away fro a while, but once in place, rebooting has no effect.
So it was back to the drawing board again. I did a load more stability checks, and this time Orthos suggested that the basic overclock wasn't stable. So I throttled back a 100MHz to 3.5Ghz and tried again. Much tweaking later and I could run small or blend for hours without issue. But the gfx problem was unchanged. I did then discover another interesting issue: Oblivion was back to CTDs - an old Bethesda favourite. Eventually I tracked this down to the RAM settings, and managed to get Oblivion to run happily for two hours without a problem. Next night the stupid rig wouldn't even play UT2004 - hardly a stretch for this system. Instead it would lock up as soon as the "Way it's meant to be played" logo appeared. Again, it's the graphics seizing, not the rig as a whole. Even AvP2, a game which came out in 2001, caused the slide show after ten minutes.
My new theory is that this is a slow deterioration of the RAM. It wouldn't be the first time I've had RAM play nicely for a coupe of months and then go ****-up - my old DFI SLI-D borked a gig of perfectly good RAM. Again, a perfectly stable o/c started to go wrong, and the RAM had to be run slower and slower to keep the rig stable. Except here running the RAM slower hasn't helped much - raising the voltage has been more effective.
It seems that something overheats, but only slowly. That overheating is largely independent of workload but simply takes a set time to happen. Once overheated, it takes a while to cool down again. Not entirely sure it's the RAM though. Ho hum.
Guess it means more ****ing playing around to avoid forking out for more RAM. Or at least to pin it absolutely on the stuff (OCZ PC8500)
M
BFG RMA'd the card saying they thought that the problem was overheating video RAM. I don't believe that it was, but it's generous of them anyway. The replacement card arrived two days after I sent them mine, so their service is good.
Guess what: either the same problem is there, or else its junior cousin. The system doesn't do a full lockup, but instead turns into a one-frame-every-twenty-seconds slide show. I can sometimes quit out of it, but not always. As before, if left off for an hour the problem stayed away fro a while, but once in place, rebooting has no effect.
So it was back to the drawing board again. I did a load more stability checks, and this time Orthos suggested that the basic overclock wasn't stable. So I throttled back a 100MHz to 3.5Ghz and tried again. Much tweaking later and I could run small or blend for hours without issue. But the gfx problem was unchanged. I did then discover another interesting issue: Oblivion was back to CTDs - an old Bethesda favourite. Eventually I tracked this down to the RAM settings, and managed to get Oblivion to run happily for two hours without a problem. Next night the stupid rig wouldn't even play UT2004 - hardly a stretch for this system. Instead it would lock up as soon as the "Way it's meant to be played" logo appeared. Again, it's the graphics seizing, not the rig as a whole. Even AvP2, a game which came out in 2001, caused the slide show after ten minutes.
My new theory is that this is a slow deterioration of the RAM. It wouldn't be the first time I've had RAM play nicely for a coupe of months and then go ****-up - my old DFI SLI-D borked a gig of perfectly good RAM. Again, a perfectly stable o/c started to go wrong, and the RAM had to be run slower and slower to keep the rig stable. Except here running the RAM slower hasn't helped much - raising the voltage has been more effective.
It seems that something overheats, but only slowly. That overheating is largely independent of workload but simply takes a set time to happen. Once overheated, it takes a while to cool down again. Not entirely sure it's the RAM though. Ho hum.
Guess it means more ****ing playing around to avoid forking out for more RAM. Or at least to pin it absolutely on the stuff (OCZ PC8500)
M
There's another thread somewhere where I said that BFG agreed to RMA the card, saying they suspected the video RAM was overheating. The new card played some odd tricks, but not the same ones. MY eventual conclusion was a faulty motherboard as well as or instead of a faulty 8800. The motherboard has since died completely.
M
M
Hi Meridian,
Just read this thread with interest. Just to let you know I bought the BFG880gtx OC2 which is factory clocked at 626 core 1000mhz ram. Right from the start I got lock ups in 3D games after a period of time. At that time I was palying a lot of EQ2. Sometimes it would play for hours without a lock up, sometimes after 5 mins. I blamed Nvidia's drivers having moved from ATI with zero problems.
I started playing AOE3 again and this would crash all the time. AgaIN i blamed drivers ETC. By now I had had the card over a month so it was too late to swap it straight to OC.
Anyhow to get on with my tale. I installed ATITOOL and clocked down the memory fo the card from 1000mhz to 975 mhz and guess what no crashes! I can run the ATI spinning cube and leave the pc with it set at 1000mhz and it locks in less than 16mins. Altho the PC hasnt crashedjust the 3D cube thingy is all corrupt and locked. When I set to 975mhz it runs all night and is stable.
Obviously my problem is the video ram. I emailed BFG last week but they havnt even acknowleged my email. Did you have to send your card back to the ~States for replacememnt? ALso have you tried lowering the settings of the card (core and ram) to see if this stops the locks? I think BFG are on dodgy ground with their factory overclocks not testing them for long enough
Cheers
Deks
Just read this thread with interest. Just to let you know I bought the BFG880gtx OC2 which is factory clocked at 626 core 1000mhz ram. Right from the start I got lock ups in 3D games after a period of time. At that time I was palying a lot of EQ2. Sometimes it would play for hours without a lock up, sometimes after 5 mins. I blamed Nvidia's drivers having moved from ATI with zero problems.
I started playing AOE3 again and this would crash all the time. AgaIN i blamed drivers ETC. By now I had had the card over a month so it was too late to swap it straight to OC.
Anyhow to get on with my tale. I installed ATITOOL and clocked down the memory fo the card from 1000mhz to 975 mhz and guess what no crashes! I can run the ATI spinning cube and leave the pc with it set at 1000mhz and it locks in less than 16mins. Altho the PC hasnt crashedjust the 3D cube thingy is all corrupt and locked. When I set to 975mhz it runs all night and is stable.
Obviously my problem is the video ram. I emailed BFG last week but they havnt even acknowleged my email. Did you have to send your card back to the ~States for replacememnt? ALso have you tried lowering the settings of the card (core and ram) to see if this stops the locks? I think BFG are on dodgy ground with their factory overclocks not testing them for long enough
Cheers
Deks
Soldato
- Joined
- 11 May 2006
- Posts
- 5,786
Have you replaced the motherboard yet?
I had a similiar problem once (on an old AGP rig) but it displayed the same inconsistent symptoms. I was on the verge of chucking away a perfectly good graphics card when I decided to replace the motherboard. Everything was sorted after that and the graphics card is still running in a spare machine I have.
Motherboard problems are not nice, thats for sure!
Deks
Yes, I tried underclocking the card - a Google search threw that up as a possible solution. All it did was delay the inevitable. RMA was to BFG's UK office, although the initial correspondence was via the US.
titaniumx3
Yes - it's running without issue on an Asus P5K-E WiFi AP.
Yes, I tried underclocking the card - a Google search threw that up as a possible solution. All it did was delay the inevitable. RMA was to BFG's UK office, although the initial correspondence was via the US.
titaniumx3
Yes - it's running without issue on an Asus P5K-E WiFi AP.
Try stock CPU and RAM and see where that takes you?. Please do it as my m8 had the same attitude as you saying that his CPU and RAM were fine with his E6600 and his SLI'd 8800GTX's. All at stock and not one crash in 5 days now so it's maybe worth a bash. If reverting back to stock speeds does help then it's up to you to overclock slowly and surely.
Meridian,
This sounds all too familiar to me. Several months back I had the same problem. I ended up replacing all components except the hard drives. I initially though the RAM was fine (until, when I replaced the CPU, somehow managed to kill the memory). After that, no lockups during any video animations etc even after 20-odd hours of running. I had throught it was the CPU itself. I had replaced my 6600 C2D with the Quad 6600.
I had also replaced the OCZ PC-8500 ram with a spare set.
The problem has come back again.
My rig consists of:
C2D Q6600
Asus Striker Extreme
XFX 8800GTX
OCZ PC-8500 (2x1Gig sticks)
Creative X-Fi F@tality soundcard
WD Caviar 250Gb SATA-2 drives (4)
Jeantech Storm 700W PSU
Two items similar to yours, the ram and the soundcard. Hard to think the sound is the issue as I can play mp3s and sound files without hitches.
Everything runs at stock speeds, no OC anywhere. I upped the VRAM voltage (as per OCZ's support suggestion) to 2.3v to get the ram to run at it's advertised speed, but no other changes.
Memtest works without errors, but I'm now losing a bit of confidence there. I'll replace the ram in a short while, after the next lockup, and post back the result, hopefully that'll help both your problem and mine.
Spadge.
This sounds all too familiar to me. Several months back I had the same problem. I ended up replacing all components except the hard drives. I initially though the RAM was fine (until, when I replaced the CPU, somehow managed to kill the memory). After that, no lockups during any video animations etc even after 20-odd hours of running. I had throught it was the CPU itself. I had replaced my 6600 C2D with the Quad 6600.
I had also replaced the OCZ PC-8500 ram with a spare set.
The problem has come back again.
My rig consists of:
C2D Q6600
Asus Striker Extreme
XFX 8800GTX
OCZ PC-8500 (2x1Gig sticks)
Creative X-Fi F@tality soundcard
WD Caviar 250Gb SATA-2 drives (4)
Jeantech Storm 700W PSU
Two items similar to yours, the ram and the soundcard. Hard to think the sound is the issue as I can play mp3s and sound files without hitches.
Everything runs at stock speeds, no OC anywhere. I upped the VRAM voltage (as per OCZ's support suggestion) to 2.3v to get the ram to run at it's advertised speed, but no other changes.
Memtest works without errors, but I'm now losing a bit of confidence there. I'll replace the ram in a short while, after the next lockup, and post back the result, hopefully that'll help both your problem and mine.
Spadge.
Maybe something is wrong with BFG cards at the moment, my new BFG 8800GT doesn't seem to want to play nice with all my games
Out of interest, what is it doing? Or not doing?
M
See here http://forums.overclockers.co.uk/showthread.php?t=17802295
I just played some ET:QW and it looks sweet as, no missing textures and no problems with it loading.
Ok the machine locked up last night while playing WoW. Managed to get back to the desktop before it completely died, so could close everything down tidily and power the machine off. I then replaced the ram, and it ran fine for four hours after that.
I left the PC on overnight, just idling (well, trying to download a new WoW patch, which didn't happen.
).
This morning, played some video, and some Sims2 for forty minutes, still working fine so far.
So, either a heat issue with the OCZ ram, or the ram itself is indeed slowly dying. I'm hoping the former as OCUK must be getting fed up with me sending back faulty OCZ sticks.
I'll see if I can get some sort of fan cooling onto the memory, without interfering with the CPU and chipset cooling on this board. I'll post the results if I get that far.
Spadge.
I left the PC on overnight, just idling (well, trying to download a new WoW patch, which didn't happen.
).This morning, played some video, and some Sims2 for forty minutes, still working fine so far.
So, either a heat issue with the OCZ ram, or the ram itself is indeed slowly dying. I'm hoping the former as OCUK must be getting fed up with me sending back faulty OCZ sticks.
I'll see if I can get some sort of fan cooling onto the memory, without interfering with the CPU and chipset cooling on this board. I'll post the results if I get that far.
Spadge.
Just a quick update, although I haven't been able to obtain suitable cooling for the memory, the PC's been running solid since the last post. I'm convinced there's something on the OCZ memory that breaks down slowly over a period of time - possibly the thermal transfer material? Meridian, try to get some active cooling on the memory in your PC if possible, or replace the memory if you can.
Cheers for now,
Spadge.
Cheers for now,
Spadge.
Just a quick upgrade, I finally got round to returning my card to BFG. I sent it on Monday and received a replacement 3 days later!!! the new card works perfect with no crashing after 1 hours testing.
I can't praise BFG's after sales service highly enough
Deks
edit. I forgot to mention, you return it to Wales (uk) which is another added bonus
I can't praise BFG's after sales service highly enough
Deks
edit. I forgot to mention, you return it to Wales (uk) which is another added bonus