I'm having an issue lately with my monitor (Analog LCD) appearing to go out of sync when starting or running an OpenGL application. This usually occurs when I first start the app but today I was able to play UT2004 for 3-minutes before this happened. I'm running an GeForce FX 5700LE O/C'd to 370/425 but the problem persists even when running at default clocks from a cool state. Using latest nvidia drivers from the repos and everything worked just a few days ago. I'm not sure what's going on here but I just wanted to see if anyone else was having a similar issue and if so, how you fixed it.
FYI: All my libs are linked correctly and direct rendering is enabled as it should be. This occurs with any 3D app from UT2004 to a GL hack for Xscreensaver to glxgears. There is no way out of this state short of pulling the plug. It seems that the monitor is out of sync but I think the whole system is hard locked. Thanks.
Does this happen in window or full screen modes? Do these apps need to change the screen resolution when you start them?
Occurs in either windows or fullscreen. It doesn't make a difference. The resolution for the programs i've tried is the same as my normal resolution (1024x768x24) so I guess that answer here is no.
sounds like the death of some hardware coming
I am not your friend
I have the same issue with GL apps on Linux. My video card is crapped out though, so I know where the issue came from.
It's possible that your OC-ing killed the GPU. :?
I'm starting to think you're right. I can get it to play a game now but it doesn't last long before it kills over.
[...] I can get it to play a game now but it doesn't last long before it kills over.
Try taking your case covers off and let it breathe "air naturale". Move some of your PCI stuff in other slots further away from the AGP, giving it a nice "channel" of "separation". See if the extra cooling helps it "play" longer. If so, maybe you've overheated it and do need another one. Be sure to set the OC'ing back to normal spec speeds too. And while you have the case off, take the card out and re-seat it "hard", without cracking any motherboard fab of course.
Also, try lowering your AGP rate from "8x" to "4x" to "2x" or whatever. You may be having motherboard problems instead. You can do that with the NVIDA driver, but I don't remember how, or maybe it's easier in your BIOS.
Just for kicks, turn on all the "sync to monitor" options with the NVIDIA setup GUI util. I forgot what it's called, "nvidia-settings" maybe? Be sure to hit "quit", not close out the window.
I think they have options in there like "sync to VBlank", "sync to blitter", and one or two others. Or, it's quite possible you may have some of those turned on accidentally, so try turning them off.
It's also quite possible that maybe you've just recently updated your drivers in both Windows and Linux, and your card just don't like them. Try 1 revision down and see if that helps.
Well, I jerked the card from my machine and slapped it into my roommate's machine and it works great. He can play UT2004 for hours on it and not a single issue whatsoever. I have experienced something of an interesting issue myself in today as well--as the power supply +5V pins on the standard ATX plug (I don't have a connector for the 2x2 one) burned through the plastic of the plug and started to melt the motherboard plug. I'm not sure what the cause is but I'm pretty sure that the motherboard and supply are dead. I did an emergency newegg order this morning for a new power supply (Thermaltake 480W) and motherboard (Albatron KX18DS PRO) because I need this system for weekend work. Hopefully I'll get these tomorrow afternoon and can do an afternoon-night build session. Still, I wish the source of my burnout were known. Makes me feel unconfortable with the new parts. I did notice that the new board doesn't have the 2x2 pin power connector either. Someone told me that most AMD boards don't have this, is true? Why would only the +5V pins burn out? I was using an Athlon XP-M @ 3200+ (2.28Ghz) on 1.625V Vcore which is acctually less voltage than the desktop 3200+ requires standard and a Chaintech GeForce FX 5700LE running at 50Mhz or so below non-Le specs. My old board (Epox 8RDA+) of course didn't detect the CPU type correctly but I never attributed any problems this and for a long time never had any issue with this configuration (used for two weeks with no issues). Anyway, sorry for the long ass post but this is driving me crazy. Thanks.
iBertus, from what I remember, the extra 4-pin connector is to supply additional current through the 12V side. I don't believe AMD motherboards use them, but I think a few did like the XP's and 64s. If you do have a 4-pin connector on an AMD or Intel board, always use it since mobo designs direct current demand by priority. I think you said you didnt have the connector on your mobo, but you did on the PS?
In that case, your AGP under heavy load will draw the current from where it can get it. I think most motherboards will tap the 12v side of your PS for drives, fans, and such. When you overclock, the extra oscillations pull in more eV's at a higher rate. It's quite possible that was what you experienced on your +5V side, since your 12v couldn't provide the difference or was tapped out already. I think most mobo designs pull from the 5/12 (in that order) for extra AGP demand.
So, even though you have a nice hefty PS with 2000 Watts, check the specs on the 3/5/12. During intensive video card demand, you can pull 3-5 Amps off your 12 -> 30-60W. AGP slots only supply ~40(?) max per design specs. So, you're talking about 30-100W alone that your 3D card may require, and that's for hardware 2-3 years old now. I paid $120 for my Enermax 3 years ago. I checked the specs on their PS sheet and matched them up with my components to make sure. I never did overclock and got rid of my "dualy" rig altoghether. But, most people overlook the importance of spending moree than $20-$40 on a PS, especially when they're running 6-fans, multi-I/O devices, and game a lot.
I had three 80mm fans and one 80mm CPU fan running plus on optical drive and one HDD. Oh, forgot the one 3.5" FD disk drive. My old supply was an Antec SmartPower 350W supply w/ dual fan design. When I finally build an AMD64 rig I'll make sure to get a high quality supply. My last board didn't have the second power connector but my PS did. Maybe now I understand why some video cards have a molex connector attached to them (maybe more should). Do the fan headers on the mobo get power from the 5V? I did have the 80mm CPU fan attached to the mobo header because I didn't have an adapter with RPM readout and my board wouldn't POST without that. New parts should arrive today via FedEx overnight service so I'l post back when everything is back together.
[...]Do the fan headers on the mobo get power from the 5V? I did have the 80mm CPU fan attached to the mobo header because I didn't have an adapter with RPM readout and my board wouldn't POST without that. [...]
All modern mobo fan headers supply a 12V "tap", especially the CPU header. Better designed motherboards will not allow you to post if your CPU fan speed is too low. Depending on the resistance (from the bearings used) in a fan, some fans only need <10mA/5V to start, and other "high flow" fans >500mA/9V, but they should be all rated @12V.
I think in your cable "burnout" case, it was just a matter of an older PS, or just the rare case of bad manufacturing from a good PS manufacturer like Antec. The more expensive PS(s) have surge protection "trip" switches instead of slow burn fuses, much like the breaker box in your house. Those are far more effective and add to PS cost.
By the way, if you wish to have a little "peace of mind", burn the "Ultimate Boot CD" ISO on a CD. I gave the link somewhere in the OT forum. Basically, it has all types of system "burn in" tools, plus memory and hardware checks, etc.