Well, my new PC parts turned up, and was carefully assembled. And to cut a long story short, I have problems with what seems to be Directdraw. The main overall problem is that I'm now suffering graphical glitches. At random, either horizontal lines will pattern fill the screen, or a chequerboard pattern will fill the screen. After this happens, the mousepointer and keyboard sometimes locks up, but always the following happens... Windows XP SP3 would BSOD, with the error reported as "nvid4_disp.dll". Windows 7 will attempt to reboot the Nvidia drivers, sometimes sucessfully. The Nvidia drivers are 258.96, for both XP and Win7. XP was very unstable, horizontal lines filled the screen during install. Win7 seems more stable, but only at 1680x1050. Why I think Directdraw is at fault is that during XP's more stable time, I ran the DXinfo application, and did a test of the video stuff. Directdraw filled the screen with its usual black and white pattern, then XP promptly locked up. Shortly after this, XP BSOD'd with the "nvid4_disp" error. In XP, I tried disabling hardware acceleration, but this didn't resolve the fault. I'm currently in Win7. I've reinstalled the 258.96 drivers, but I still can't use 1920x1080 reliably. My hardware is a Palit 768MB GTX460 (not overclocked), and is connected to my monitor with the VGA cable. The CPU isn't overclocked, and is an AMD Phenom ii 940 3.0GHz thing. The motherboard is a Gigabyte GA-870A-UD3, and I've upgraded the BIOS to F3a. I've swapped the graphics card from the x16 PCI-e slot to the x4 PCI-e slot, mainly to access the CMOS jumper easily, but also to see if the different slots do anything different; they don't. I've also reset the CMOS, and set the graphics card choice setting to "PEG1" (PCI-e x4). I'm out of ideas... anyone else got any clues? Edit: I forgot to add the PSU I'm using is an Antec TP-650.
that sounds like an overheating or cooked graphics card, rather than a driver issue check the fan on the graphics card is spinning, and that all power cables are correct, and power supply is sufficient to run that card. make sure the 12v rails are sufficient.
Thanks for the reply. Yep, the fan's spinning, and the power cables are connected. My problem is that there's so many combinations of PCI-e power cables coming from the (modular) power supply. Do I use the one cable which has two PCI-e connectors on it, or do I try using one of the modular plug-in PCI-e cables instead? Do I mix the connectors... one from the modular cable, one from the connected cable? So far... so far... I've got the card sat back in the PCI-e x16 slot, with one power connector coming from a modular connection, the other connector is plugged in to a molex convertor that came with the card. I've also set the BIOS to the "PEG" setting, as opposed to "PCI" or "PEG1", and I'm using the VGA cable to the monitor. It had a little colour flicker, but nothing nearly as bad as I've seen, and that's not happened since (my guess is that something somewhere didn't like the case being tilted slightly.. the joys of working in a confined space). Airflow isn't a problem as yet, the HAF922 has its stock fans connected to a molex connection, so they're running full speed (and they're nowhere *near* as loud as my last PC). A couple more questions. Regarding the fan on the graphics card. The Nvidia drivers installed something called "vtune". In this Vtune thing, there's a function to control the fan speed... that's current set at automatic. Would it be detrimental to the system if I set this to manual, and bumped the setting up to 100%? Also, could not enough CPU thermal paste also be a cause of the graphical glitches? Yes I put some on, and I made damn sure I didn't smother the CPU with an inch-thick layer... but I'm wondering if me putting not quite enough on would be part of the problem. Again, thanks for the help offered. Time to see if this seeming-stable PC is actually going to behave today.
Setting your fan to 100% shouldn't be a problem for testing purposes, except for the noise it'll produce. After you get it stable I would set it to auto again though, also for the noise levels. As for the power cables, you could try with 2x 6 pin PCI-E modular cables. If this fails I'd use the one from the non-modular cables, and another 6 pin PCI-E modular cable. If it works you could attempt to go back to the non-modular 2x 6 pin PCI-E connectors. You should find a solution first, then worry about the cabling. So yea, all about just testing what works. If none of this works it might just be faulty..
if your using the one cable, with 2 connectors at the moment, then i would swap for 2 seperate cables to see if that makes a difference. if its a modular power supply, make sure you are using the correct sockets from the power supply, they are normally coloured differently. thermal paste is important, you want i thin layer, but not 2 thin that it doesn't make contact with both surfaces. the way i do it, is to squeeze a small amount of paste onto chip, then place cooler over it and press, and twist slightly, check that its spread nicely,without spilling over edges, apply more if needed. the amount you squeeze out depends on the size of the chip, very small chip about a quarter of a pea should be enough, half a pea for intel 775 size chip, seems to be enough. fan speed if you set it to 100% the only detrimental thing maybe the turbine like noise coming from pc, would become annoying, and maybe the fan will wear out a bit quicker. the cooler you can keep the card the better to be honest
I have done, as the problem reappeared shortly after I tried playing a video full-screen. I've also put the card back into the PCI-e x4 slot, and I've set the desktop back down to 1680x1050. Once again, so far... so good. Yeah, I'll have to read up on the manual on the Antec website... the sheet I got with the PSU was very useless. Hmm... maybe the old "grain of rice" amount is... urgh. Well, I do have some heatsink gunk cleaner, and it's not as if the H/S is as awkward to attach as they used to be. The odd thing was, the last time I set the fan speed manually (and heard the fan scream), the graphic glitches returned. To me, that says that it's a power issue, so I've swapped the molex-converted connector out for one of the fixed cable PCI-e connectors. The other odd thing was, only when I set the fan speed manually, did I hear the fan speed up. When I tried Quake 1 Darkplaces, the game just glitched... the fan itself didn't speed up.
to me you have 1 of 2 problems. either faulty GPU. RMA it. or faulty mobo. RMA it. test the GPU in another PC if it still glitches its faulty if it doesnt its probly the board.
I'm coming to that conclusion, sadly. Another thing that might help with the diagnosis, is that sometimes it'll show the horizontal lines at the post screen, but it's never shown the problem whilst in the actual BIOS menu screens.
It really sounds like a power issue to me too... do you have the option of testing the GPU in another system? and/or the option of switching the PSU I can't really see how it could be a faulty mobo.. Could possibly just be you, since (I think) you've booted it far more often then you went in the BIOS, and since you said it would sometimes appear it post screen, you might've been "lucky" and not seen it in BIOS yet.
it can quite easily be a faulty mobo since you plug the GPU into it it could be faulty PCI-e connectors could be faulty northbridge dodgy soldering etc Only way your gunna find out is testing the components seperately by swapping them out. test a different GPU test a different PSU if the problem is still present then you have a motherboard issue
Well, I tried the graphics card in my girlfriends PC last night... it showed exactly the same problem.
