Unstable system after installing gtx275; cooked something?

D

dfidler

n00b
Joined
Nov 14, 2009
Messages
11
Hi All,

Summary:

I installed a gtx275 and got some serious instability;
  • intermittent locks
  • spontaneous reboots
  • BSOD: IRQL_NOT_LESS_OR_EQUAL
  • System Event Viewer says "the nvidia display driver service has reported an invalid current state 32" (which google says is a bug in the 19x.xx drivers), etc.
  • After reboot, onboard NIC wasn't detectable; warm boot fixed it
  • chkdsk found errors on disk after one spontaneous reboot (not surprising if it was writing to disk when the system spontaneously rebooted)

Frustrated with this problems I removed the gtx275 and replaced it with my time (and game) proven 8600gts... but the stability issues persist!

Aaaargh!!!

The rest of the story....

With the install of the gtx275 I was getting all of the issues listed above. To fix it I tried:
  • upgrading drivers to 191.07 (bad idea; buggy)
  • downgrading drivers to 186.18 (uinstalled 191.07 first)
  • underclocking using rivatuner
  • stress testing with FurMark (4 hours, no crashes)
  • open case to air + open windows to winter temps
  • disabled windows 7 "background slide show"; this reduced frequency of crashes to about once per 30-120 mins.

Through all of this, temps were well within norms: (degC idle-loaded)
  • CPU: 37-47
  • GPU: 45-84
  • Mobo: 38-45

So I eventually gave up on the gtx275 and have it packed up for RMA. But I'm still getting the crashes after plugging my 8600gts in. COD4 (which never used to crash) is now barely working at all.

So I've tried:
  • uninstalling rivatuner
  • uninstalling nvidia drivers
  • installing GPU drivers from windows update (system was stable with these prior to this whole fiasco)
  • memtest - got two errors in 1018% coverate (whatever that means)
  • pulling RAM pairs and running through memtest (no issues so far)

Strangely, the system hasn't crashed while running memtest for many many many hours. It seems that the system will only lock when I'm actually using it (if that makes any sense). ie - opening browser tabs, watching a video, loading a program, playing a game, wiggling mouse to wake up the monitor after being ilde for a while, etc.

This system has been stable for 2 years. The only changes were:
  • Install W7 64bit 2mo ago
  • Install 2x2GB RAM sticks2 mo ago (total of 8G)
  • Disabling windows page file (so only RAM is used); 2 mo ago

Note: I use many memory hungry apps; photoshop, lightwave, COD4, etc so I would have expected to have seen issues from bad memory a long time ago.

I'm starting to wonder if the gtx didn't bake my psu or something on my mobo; is this at all likely? Anyone know how I can test these components? And are there any really good stability tests for just the GPU or was furmark the right way to go?

System:
OS: Windows 7 64bit
CPU: Intel Q6600 G0
Cooler Tuniq Tower 120
Mobo: ASUS P5N32-E SLI (i680 chipset)
RAM: 4x2GB DDR2 Corsair (XMS2-6400, 4x2GB, 5-5-5-18)
PSU: Thermaltake 750W (750 AP- W0117)

Thanks, in advance, for any help/advice you can provide.

Cheers,
Dave.
 
If it were me, my next step would be a full reinstall of W7. (Thankfully W7 is waaay faster than XP when installing.)

Eliminate all software possibilities and then look to hardware. At least it's what I would do.
 
@Little_4nic8er - I was thinking that too but am trying to avoid it if I can. I have dozens of programs installed and just re-installing those will take days. Though now that I think about it, i could always install to a new disk for a quick. :p

@Lorien - I don't have an NF200 chipset, but the symptoms that they describe do sound similar. However, I was experiencing crashes even before installing the 191.07 driver. But thanks for the heads up; I'll try the reinstall per above. But damn is that thread is worrying! :(

As an aside, I should note that I have another one of these ASUS boards... or rather I did. I just replaced the one in my HTPC because it would, periodically, refuse to boot unless I cleared the CMOS or changed some hardware (ie - remove PCI card, boot, power off, replace PCI card, boot; voila).

I must admit that I'm seriously considering purging ASUS from my life forever. I've only ever had one board (the current one that I'm having issues with) that has not had some problem with it from day one (and I've had many... faulty memory channel, faulty AGP slot, boot issues, etc) where as I had an MSI that worked flawlessly (for 3 years before an upgrade) and a gigabyte that worked well for 6mo went dark.
 
Actually, that BSOD error you see there is a hardware issue. It usually indicates issues accessing memory somewhere, but as a general rule says theres something very unstable about the hardware you're using (look at the error... IRQ... hardware uses IRQs for the OS to access it through the HAL). If the GTX 275 is the only thing you changed about your setup, RMA it. The last time I encountered this issue, it was the Crucial BallistiX RAM we had in the machine. Switching to 4GB of regular Crucial RAM over the 2GB of BallistiX got rid of the issue.

10 minutes of googling and reading can save you several frustrating posts by forum 'gurus' ;)
 
@LAiN_OfTheWired - The problem is that the gtx isn't even in my system any more and I'm still getting system instability with every day tasks like web browsing, etc. Granted, the period between crashes is greatly reduced.

That's why I'm testing the memory modules (I'm currently scanning the second pair; memtest: ~520% across three threads; no errors so far).

I am inclined to agree, though, that the gtx is part of the problem. It crashed my system, the first time within about 5 minutes of playing COD4:MW2. I was able to reduce the frequency of the crashes overall, but not eliminate them.

As for googling, I did read a lot of threads re: IRQL_* saying that it's a hardware issue, but most said that it would be either the GPU, memory or heat issue. I've ruled out the heat issue, and as the GPU is no longer in there, the GPU has also been "ruled out".

That leaves me with bad memory or bad mobo. So far, the memory is coming up clean... which is making me wonder about my mobo (or even my PSU).

Google has told me that there have been a number of problems with the P5N32-E SLI boards; many to do with adding a second pair of RAM in them, so I'm really starting to think that's what's biting me.

Mental note to self, I went off ECC RAM 5 years back because of cost (I never used to buy anything but), but after this hassle, I think I might become an ECC fan-boy again.
 
@Lorien - I found out that that NF200 chipset is basically just a i680 + PCI-e v2.0 support. So maybe I spoke too soon by saying that the 191.07 driver issue shouldn't bite me.

I was just discussing all of this with my wife and she asked, "So if the driver did cook your motherboard, is nVidia going to buy you a new one?"

I wish! lol! :)
 
You said in your one post that you got 3 errors when running memtest..... I would say it is definitely a memory error.

Memtest 86+(the one you boot to from a CD) doesn't put very much stress on the RAM. If you get even one error with memtest86+, the RAM is bad.

A better one to use that actually puts stress on the RAM and shows up errors a lot better is HCI memtest. http://hcidesign.com/memtest/

Also, have you tried reseating your CPU?

What about resetting the BIOS?

It could also be that your PSU had a marginal part in it and the GTX275 just pushed it over the edge.
 
This is indeed strange to hear.
My son has an ASUS P5N32-E SLI board (680i) running a Q6600 and 4 GB of Corsair Dominator, has been for over a year solid. No changes.

I had an EVGA GTX 260 in his box, and all has been well since Jan 09.

I upgraded my GPUs to ATI 5870s, and gave my son one of my GTX 285s.
The GPU is a BFG GTX 285 OC model that ran fine in my computer, in SLI, with 190 drivers ON an evga Classified with the NF 200 chip.

The minute the GPU was put in my sons rig it started to get flaky, COD MW2, Borderlands, CS:S, you name it all crashed at random.

I swept everything clean before I installed the 285. I used every driver available, 190s,191s and the beta 195......all crashed.....with messages of NV4.dll......I couldn't find a thing wrong, all temps normal,memtest normal,Orthos stable x 4 cores x 24 hours.

My son deleted everything he didn't need and now the computer is solid as a rock again......
He played COD MW2 practically all day yesterday using the 191.07 WHQL drivers with no problems.

System:
ASUS P5N32-E SLI
Corsair Dominator 4 GB 1066 DDR2
Q6600 @ 3.1 GHz
Corsair TX-750 W
BFG GTX 285 running at 675 MHz......temps normal per MSI Afterburner with fan at 70%.
 
Last edited:
@cyclone3d - I'm using HCI's memtest for all of this.

I haven't reseated my CPU; shouldn't have to with 2 years running stable. I haven't reset my bios, will give that a shot.

re: PSU, I'd thought of that too. I've got a Coolermaster 700W downstairs in my HTPC that I can pull (and probably will after memtest is finished).

Going back for a sec, I've just thought about something. The first memory scan was done in W7, regular boot up. All subsequent scans have been run in safe mode (to ensure that a max amount of memory was available to scan). Then net effect of this is that most of my drivers probably haven't loaded.

I might just work in safe mode (with networking support) see see if the crashes persist. If they don't then it's likely a driver isue.
 
@magoo - I missed your message the first time I read through this... that is interesting indeed.

When you say that your son "deleted everything he didn't need", what do you mean by that? Do you mean that he uninstalled drivers or other programs?
 
If youve got memory errors, and taking out the gtx 275 makes things more stable but it will still crash, there are 2 things to look at: the motherboard and the power supply. It's possible the PSU was simply overworked with the GTX 275 and either is starting to flake out or there was something power-wise that could have started to bugger something on the mobo like the northbridge, memory slots, or memory itself. I was lucky enough I didn't have to go that far, but an absolute fresh start would be the preferred way to go before replacing anything.

pull the drive if you have a spare you can install windows fresh on, update the BIOS on the mobo, reseat all of the hardware, and give it a go like it were a newly built PC. If everything is golden, obviously its software. If it is not, start replacing things. Since the memory shot out some errors during testing, I'd start with that.
 
I may have missed this but what type of gpu did you have in there before? Did you uninstall the nvidia drivers prior to putting in the 275?
 
dfidler said:
@magoo - I missed your message the first time I read through this... that is interesting indeed.

When you say that your son "deleted everything he didn't need", what do you mean by that? Do you mean that he uninstalled drivers or other programs?
Click to expand...

He deleted unused programs.
We kept any drivers he uses/used intact, such as chipset,audio,and some utilities, such as Adobe and MSI Afterburner and ESET Smart Security.

He had a ton of junk like download managers that we got rid of, and a number of games he hasn't used in a long while.

After that everything was fine, even with the GTX 285 at Factory OCs of 675/1250.....I'm not sure where the magic is?
 
@deadlift1 - was an 8600gts; didn't bother to uninstall drivers before updating to 191.07.

@magoo - um... weird. :(
 
You must log in or register to reply here.
Back
Top