Graphic woes AKA light speed

[Iceblade]

Okay, I have been having some issues where my graphics card will shut down and be restarted by Vista in several somewhat graphic intensive games (including Standoff).

Games like Doom3 (and the DarkMod modification for Doom3) would play for a few minutes than black screen and freeze. Also Thief Deadly Shadows would flash a black screen a few times and eventually have the same black screen freeze as with Doom3. Every time there was a black screen, I would get a message from Vista telling me that the video card was successfully restarted.

(Doom 3 is a Steam game while TDS being played from a CD)

I've been troubleshooting off and on for months now (ever since mid-October when I really started to detect a significant problem - the DarkMod release basically).

Eventually got to the point where I thought it might be insufficient power going into my Nvidia GTX 285 (550 Watt power supply but the power going into the card was either 17 Amps or 33 Amps - not sure if both 12 V rails were connected to it or not). Any case, new power supply installed a day or so ago, and not much if any improvement with the crashes.

Next I decided to see if Standoff would have similar issues and it did. Tried the high-res patched version of Secret Ops and with max resolution and DirectX 16bit, I get the attached mess in the briefing and similar graphical problems in the game. Note these problems are gone when playing in glide, also I didn't have any of these particular graphics issues problems in Standoff (which obviously is set to run glide only), but the game still crashed as before with Doom3.

So some of the other things I've tried with limited success (no success in the games mentioned above):
different versions of Nvidia drivers including the latest
removed KB952287 windows hotfix
expanded the nvlddmkm.sy_ and replaced nvlddmkm.sys in the system32 directory
checked the video card with a stress test and all values returned were good or excellent
updated directX to Feb2010 version

So any thoughts as to the probable culprit. I have a few ideas, but I would like to hear what some others might think as well.

Specs:
Vista 64-bit OS (premium)
BFG Nvidia GTX 285 (1 GB and not overclocked)
BFG EX Series 1000 Watt power supply (two 12V rails of 36 Amp apiece going into video card)
RAM: 6 GB DDR2 dual channel
Intel Core 2 Quad Processor
Intel motherboard (not sure which series though) with integrated sound card

Machine is mostly a store bought Gateway excluding both power supplies that have been used, the graphics card, and a secondary hard drive (a 7-10 year old Maxtor that I currently only play Secret Ops and Standoff on).

Attached are pictures of the mess observed in the briefing (can't really explain but it seems like the prior frames are left behind as the camera and ships go through the briefing) and the issues with the star dust trailing behind giving an almost light-speed quality to this POV.

 

[Claymore]

I don't know anything about driver settings in Nvidia drivers. You should try turning things off and on in it. If its just a setting in there you should be able to figure it out. I suspect the problem is here. What does dxdiag say about your card? It looks like 2D settings are whack.

I also kinda wonder if something in bios is wrong. Like just a PCI Express setting somewhere in it.

Gateway boards are proprietary, so their website should have different bios releases.

I had a rage pro years ago that would start running Need For Speed 3 in "wire frame mode" and i never figured out what made it do it.

I also know that you can flash graphics cards. But this is much more risky then flashing your mother board. You have to know what brand your graphics card is, what brand of ram (also the ram chips speed and size) is installed in your card. Not suggested at all.

Basically,
1.) Completely uninstall/ re install graphics card drivers. There should be a guide somewhere on the net that specifically describes how to do this with Nvidia cards.
2.) Install latest DirectX
3.) Check/Change Settings in Nvidia's manager
4.) Check Voltages in Bios
5.) Check PCI Express Settings in Bios
6.) Flash Motherboards bios (IMPORTANT: Only get bio's from Gateway as their boards are proprietary)
7.) All else fails and at the risk of bricking a graphics card; carefully flash your graphics card.

 

[AD]

It could be that the fan is dead on your graphics card and it's overheating... My old ATI used to do something similar if I boosted the graphics settings with it's overdrive feature. There was a feature to keep the PC from entirely crashing by reseting the video card.

 

[Claymore]

Ad makes more sense than me. Sorry for being so complicated. If your card is over heating you could use Riva tuner and set your fan higher to keep temps down. I'm also assuming your fan is still alive.

 

[Iceblade]

Oh, forgot to mention: already checked for overheating and I've turned the fan speed up to max and I've even underclocked my video card with no success. Tinkered with the settings also and set it for max performance with only a minute or two of additional playtime to show for it.

 

[AD]

Oh, forgot to mention: already checked for overheating and I've turned the fan speed up to max and I've even underclocked my video card with no success. Tinkered with the settings also and set it for max performance with only a minute or two of additional playtime to show for it.

Can you tell (when the machine is on) if the fan on the GPU is actually running? I assume it's probably not the issue, though pLaying your games with the case open for a minute so you can put your hand by it will tell you in a hurry if it's working or not. It's also possible there might be a different defect or conflict in the card.

If you can't find any issues with the card at a cursory glance, do you have a separate video card you could swap out and test to see if you still get the same issues with the games crashing?

I have Vista 64bit on my machine as well. I used to get system-wide crashes when I first got the machine. I figured it was connected to the video drivers. I have a 9800GTX or something like that. Anyway, as soon as I updated the video drivers, the crashes went away and my machine has pretty much never crashed once since. My current driver version is 8.16.11.9107

 

[Mace]

You have made sure the Nvidia PhysX component is de-installed from the configuration screen/software menu right?

 

[Silanda]

Iceblade, what temperature is your card reading under load? Anyways, I'd follow Claymore's advice except I wouldn't flash the motherboard's BIOS unless there is a known issue with the version you currently have. Bad flashing is an easy way to brick your system, and it does happen occasionally.

When you uninstall the graphics drivers run something like driver sweeper in safe mode to clean out their remains.

 

[Iceblade]

I'll do another check of the temperatures (been months since I actually did a temperature check over the span of the crash) when I get back home later today.

The PhysX is still in the menu, but I believe it is disabled.

Already tried uninstalling drivers and the driver sweeper, but I will make a second attempt as it seems the nvlddmkm.sys was not removed last time even after driver sweeper went through and cleaned out the Nvidia components. This time though, I will be deleting any Nvidia file I find in system32.

Yeah, and doing anything with Bios beyond tinkering with the PCI-e settings is a last resort.

I really wonder though if the problem doesn't have something to do with the way the video card is installed. The design of this Gateway is very poor. Had to rearrange drives and other stuff just to get the damn card to fit to begin with. Very tight, so I don't know if the card is even using PCI-e slot 1 on the motherboard.

 

[Claymore]

I had hope just enabling/disabling all the "bells and whistles" in Nvidia's managaer would have fixed things.

Have you deleted any integrated graphics drivers that are on your machine? If you havent just be careful and not delete your chipset's drivers.

I'm doubting Omega drivers would fix your machine but you could always try them.

 

[Iceblade]

Okay, GPU temperature is at 62 degrees C several minutes after the crash event. A little hot, but nothing really too high. Rather normal temperatures for the full loads.

Tried underclocking, though I wonder if the clock settings are actually remaining this low and not reverting to higher levels.

Default core: 694
Setting core: 650
Gave a message saying do you want to keep these settings and the settings had changed to 691 core (something like that any as I'm not sure since my brother was the one making the speed changes)

The video card was definitely overclocked at the factory though, but I'm not sure whether I bought an overclocked card or not.

The result is actually getting a quicker crash, though my brother did have to do a lot of driver uninstalling and installing to get the system tools to work, so my machine was one for a good number of hours.

I'm really starting to run out of ideas. Aside from doing a complete driver clean and installing the oldest drivers I can find or trying another video card (if I can find one new enough).

 

[Silanda]

Okay, GPU temperature is at 62 degrees C several minutes after the crash event. A little hot, but nothing really too high. Rather normal temperatures for the full loads.

Yes, but running Secret Ops is not what I'd call full load. My factory OCed GTX280 (which should run hotter than your 285) only gets a bit hotter than that running Crysis with the fan on max while the game is playing. Now, I can't be certain about the GTX285, but if you have the fan at full speed, and it is that hot several minutes after a 10+ year old game has been shut down, IMHO that seems excessive.

 

[Iceblade]

I've been using the dark mod to test for the crash. SO doesn't even crash vanilla.

Btw, I just downclocked the nvidia standard clock speeds. And the temperature dropped to 52 and so far no crash after 15 minutes or so. I'm going to start testing with more involved missions for the dark mod.

Apparently it was poorly overclocked, yet I'm still not sure I brought an overclocked card.

 

[Jason_Ryock]

Have you looked at this?

http://www.incgamers.com/News/21293/

 

[Iceblade]

Not really, but I never installed those drivers and those definitely aren't the culprit.

 

[Claymore]

Have you looked at this?

http://www.incgamers.com/News/21293/

That should be front page news for this community.

I've been using the dark mod to test for the crash. SO doesn't even crash vanilla.

Btw, I just downclocked the nvidia standard clock speeds. And the temperature dropped to 52 and so far no crash after 15 minutes or so. I'm going to start testing with more involved missions for the dark mod.

Apparently it was poorly overclocked, yet I'm still not sure I brought an overclocked card.

Woo hoo it's fixed

 

[Iceblade]

Ah man, last night working fine. Today, the clock settings when back to their old values, set to the values I set yesterday and it still crashes like it did before.

What gives?

Temperature is sitting at a cool 47 (btw, I figured out why the temperature was lower yesterday, I had upped my fan speed from 14% current under automatic to 70%....ridiculous).

Basically the only real differences I see are that my computer hasn't been for very long and the current measured GPU processor clock is sigificantly lower than yesterday (799 MHz compared to some 1200 or 1300 MHz value)

 

[Silanda]

Basically the only real differences I see are that my computer hasn't been for very long and the current measured GPU processor clock is sigificantly lower than yesterday (799 MHz compared to some 1200 or 1300 MHz value)

Nvidia cards dynamically downclock when they're not really in use. 14% for the fan speed is insanely low, that will cause an overheat. Honestly, if the card isn't that old I would consider RMAing the thing.

 

[Iceblade]

Yeah, I did detect that when I played the game a little.

 

[Oggy]

If RMA isn't an option, you could perhaps try downloading nTune and overriding the fan to 100% permanently (I do this everytime I get a card, old habit from ages back), it will mean the fan won't last as long, but in all likelyhood you'll replace the card long before it wears out if you are an avid gamer.

If its a faulty fan or bad memory on the card, or any other of the 101 things that could go wrong it might be as best to just RMA it.

I have to say, I bought BFG card once a few years back, and swore I would never go with them again (loads of trouble), ever since then I have always gone XFX.