[aragond]

Folks,

I am at my f---ing wit's end. I haven't had a decent Nvidia driver since about the 270s, but now I can't play anything with anything.

Games are now just freezing. Doesn't take long either. Start, move a few metres in-game and ... stop, or stop and BZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ like a stadium full of vuvuzelas just moved into my sound card.

Thing is, it is NOT 100% consistent. Sometimes, over the past few weeks, I have been able to play for an hour -- the maximum length of time work allows me to live my life but that's another story -- without incident. But, others, freeze-city.

Is it specific games, you ask? Well, yes, but no. It seems to afflict them all at varying times and in varying ways.

Roll-back, I hear you cry! Roll-back your drivers!

Thing is, this has been happening for weeks, and the roll-backs would only roll me back to the last drivers that were busted. I am presently on 301.42 which I upgraded to from 295.73 which itself was a roll-back from 296.10, which was an upgrade from 290.36, which was an upgrade from... no, actually, I *think* that was the one I rolled-back into 285.79 and then 285.62 which was the last kinda working one.

But, I upgraded about two months ago from 285.62 because I kept getting warnings and alerts from Win7's ... the admin' log thingy* that kept telling me that every hour or so, the Nvidia drivers were crashing. So, I upgraded.

* I'm too irate to go find it again, and it's too well-hidden in the bowels of windows to want to be found.

About seven months ago, I returned my only-18-months-old me-spec'd-store-built PC because it was having what I thought were GPU issues, but they reckoned the SSD needed a firmware upgrade, known issues, blah-di-blah-di-blah. They said they tested the GPU with 3DMark or something and found no issues.

Here's my question: ... yeah, I dunno ... the first one that comes to mind is "Do Nvidia test ANY of the drivers they release, or is that our jobs?" It seems to me that it's THEIR hardware, and the driver is ONLY working on THEIR known, defined and specific hardware, so how hard could making workable drivers be? It'd be like Apple failing to write an OS version that worked properly. There is no excuse, to my mind.

But, really, Nvidia's contemptable ineptitude isn't the issue.

I guess I am really looking for any advice from any one that has had and resolved this sort of issue. Is this common, or is my computer SO unique and SO special as to simply ATTRACT these sorts of problems?

(For example, I have never, ever, ever, ever, no matter how much support from R* I get, been able to run that LA Noire on this machine. I always get to the same place and it just sits there playing menu music but presenting no menu. R* have done all-but ask me to erase my HDD and start over. Or, another example, I've never been able to play Crysis II in DX11. Freeze, burn, crash, black-out, nothing has ever worked. In fact, I struggle to think of any DX11 I've used. And this on a GeF. 580 card.)

Because, frankly, I spec'd-up and spent big and I don't really think I am getting anything close to value for money, least of all from Nvidia.

Inb4 ATI trolls and all other trolls.

[yojo2]

I had stability issues with my GTX470, though my card was very unstable when idling and surfing the Web. What solved the problem was... connecting the monitor to another DVI port. I'm dead serious.

[Exxon]

Maybe your GPU is just malfunctioning Aragorn

[aragond]

QUOTE (Exxon @ Saturday, Jul 7 2012, 20:20)

Maybe your GPU is just malfunctioning Aragorn

Yeah, 'kay, but how do I prove it?
I'm not going out and buying a new one if there's nothing wrong with it.

[Exxon]

You could download MSI Afterburner and use the MSI Kombustor that comes with it to stress the GPU.
If you see artifacts, or the same happens (crashes, BSOD or whatever) there's a big chance it's your GPU.

Or, have you got any old GPU's lieing around?
You could just buy something like a GT210 or some crappy card like that to see if your system still crashes and such.
Don't come up with that bullsh*t like you don't have enough money for that because you have a freaking GTX 580

This post has been edited by Exxon on Saturday, Jul 7 2012, 11:07

[luceberg]

I would suggest you check your ram first.
Run prime95 on blend and see if it is stable.

[Wolf68k]

Run memtest86+ http://www.memtest.org/ Let it run for several hours. Yes hours. You want it to do at least 3 to 6 passes, so let it run over night, while at work or school, go watch some movies, go spend time with your family and friends, whatever. If that passes with no errors then you're good there.

As already suggested use MSI Afterburner which comes with Kombuster as well. Watch Afterburner while running Kombuster to see if anything happen with the temps while watching Kombuster for any artifacts.


Now for my personal experience. I normally leave my system up 24/7, turn the monitors off at night. The first GTX 560Ti I got was bad but I didn't know it for a couple months. Then once a week, like clock work, whether I played games or not the drives would give me the error about it stopped and was restarted. After documenting that for a few weeks and swapping to another card for a couple to narrow the problem to the OS or the hardware I was able to conclude that the 560Ti was the problem. Contacted eVGA and told them of my tale and they agreed that it was the card and RMA'ed it.
The new card is for the most part ok, I'm not sure if the problem I'm having with it is hardware or driver but every time I update the drivers after about 1 month I start getting problems but then I swap it for a backup card (GTS450) for a few hours and then back the 560 and it's fine. My next plan when I update the drivers again I'm going to put the swap it for the 450 then install the drivers then put the 560 back in and see what happens.

This post has been edited by Wolf68k on Saturday, Jul 7 2012, 17:20

[sivispacem]

I could be a whole myriad of other issues not directly related to the GPU. Dodgy 12v power supply with lots of fluctuations? That would cause similar symptoms to the ones you describe. As would failing components in the motherboard regarding power modulation or control of PCI-E rails. The GPU itself could be a bit dodgy. For what its worth I've never had a single problem with stable release NVidia drivers- every issue I've ever had that I've suspected was drivers has eventually boiled down to a hardware issue.

[aragond]

QUOTE (Wolf68k @ Sunday, Jul 8 2012, 03:17)

Run memtest86+ http://www.memtest.org/ Let it run for several hours. Yes hours. You want it to do at least 3 to 6 passes, so let it run over night, while at work or school, go watch some movies, go spend time with your family and friends, whatever. If that passes with no errors then you're good there.

eah, good, good. Um. Hate to be a bother, but... um, how does one use a "pre-build bootable ISO" or a "bootable binary"? There doesn't seem to be instructions with either just sort-of the assumption you already know how.

Afterword: Alright, well, okay, Microsoft and WinRAR stopped fighting eachother long enough to let me burn an ISO image. I'll let y'all know if it works.

PS; The machine did something new last night: black screen and no keyboard input could undo it. The screen just wenbt black. I was typing... boom... black and no recovery. So, I still very strongly suspect that POS gpu.
Which is ironic considering I bought the expensive end of the market deliberately to get higher quality.

[MIKON8ERISBACK]

I have a slight sensation that your problem may be a defective power transformer (the thing in the back of the CPU with the fan, power switch, and 115-240 volt selector on it) overvolting your motherboard and slowly cooking your hardware to their death. A very bizzare situation happened to me with an Acer Aspire PC from 2008. Im probably wrong and your problem is more likely a driver software update that fixed what wasn't broke. I've had a couple traumatic experiences with dodgy driver software updates in the past.

[luceberg]

QUOTE (aragond @ Thursday, Jul 12 2012, 06:09)

PS; The machine did something new last night: black screen and no keyboard input could undo it. The screen just wenbt black. I was typing... boom... black and no recovery. So, I still very strongly suspect that POS gpu. Which is ironic considering I bought the expensive end of the market deliberately to get higher quality.

That is something similar to what yojo is describing above and has happened to me too. Try plugging into another dvi port. Also, make sure the connection is good. Screw in dvi port screws fully on both pc and monitor.

[Wolf68k]

So I take it you got the ISO burned to CD and were able to test the memory?

As for the black screen with no response; maybe it's not the card. Maybe it's the PSU is going out?

[aragond]

IF YOU ARE READING THIS LINE, I AM EITHER STILL TYPING OR I HAVE BEEN REBOOTED BY MY VERY FLAKY PC
Okay, well, it's been an eventful frakkin' 48 hours.
Thank you all for your persistent comments. Everyone's experiences are at least telling me sh!7 happens... a LOT.
Let me see if I can boil it down before I get another forced reboot.

QUOTE (Wolf68k @ Friday, Jul 13 2012, 01:56)

So I take it you got the ISO burned to CD and were able to test the memory?

Yes, thank you. Sorry for being a noob. After your nineteenth reboot you start getting cranky and nebbish.
MemTest ran for about 7-ish hours, not a single complaint.
No reboots either which is interesting.

QUOTE (Wolf68k @ Friday, Jul 13 2012, 01:56)

As for the black screen with no response; maybe it's not the card. Maybe it's the PSU is going out?

QUOTE (MIKON8ERISBACK @ Thursday, Jul 12 2012, 21:36)

I have a slight sensation that your problem may be a defective power transformer (the thing in the back of the CPU with the fan, power switch, and 115-240 volt selector on it)

LOL. Yeah, noob I may be, but I am familiar with power supplies.
Guys, I am starting to agree, but I am also suspecting maybe my SSD, too.

Here's what's been happening:

• Ran MemTest to brutalise my memory for seven hours. No issues.
• Shutdown, leave it be. Start machine next day, play low-intensity game like S.C.Conviction for ~40min, it runs but it runs with some glitches. On about three occasions, it "stutters" for lack of a better way to explain it. The sound stutters (in a very digital way), and the video freezes but only for parts-of-seconds and then will do it again a second later, and then a third time... and then it's fine. Game continues on okay, until the next stutter. There were about three within the first ten minutes of playing, and then the game was fine. This is the first time this stuttering has EVER happened before.
• Stop playing after 40 min, leave it still running but not played for a few hours. No issue. Start playing it again and...- BAM! Machine freezes.
• Reboot machine
• Run 3DMark11, just the Demo. BAM! It froze the vision, but the sound kept going until pretty much to the end of the first part of the demo and then just sat there. Yeah, it's the GPU, concluded I. Gotta be!
• Reset.
• Start HWmonitor. Rerun 3DMark11 demo in windowed mode, 800x600. Watch temperatures esp. on GPU. No surges at all. In fact, my cpu runs at a stress-free 30-40°C. My GPU did reach 70°C during the test but only briefly. GPU fan never got over about 53% of top-speed.
• Rerun 3DMark11 demo in windowed mode, hi-specs, 1920x1080. CPU still at 30-40°C, and while GPU did reach 80°C during the demo, only briefly. GPU fan peaked at 60% of top-speed.
• Rerun 3DMark11 Test#1 in windowed mode, 1920x1080. Temperatures ar...- BAM! Machine freezes. I decide to leave it be to see whether it ever comes back.
• PC thinks about it, and eventually reboots itself. Oh, that's new.
• ... can't recall, but I think I did no more tests, just ran Windows Apps, surfing. No issues.
• Reviewed Event Viewer. These reboots are consistently the same: there is NO MENTION of any driver failure, of a critical error, of a massive failure. The only thing you get to read is the usual pitter-patter of "The Computer Browser service entered the running state." followed by "The Computer Browser service entered the stopped state." and then a couple o' minutes gap followed by "The operating system started at system time ‎2012‎-‎07‎-‎14T07:11:31.610798400Z."
• Start Spec'Ops. I start the last chapter, something unintensive, where there's little user interaction, it's mostly cutscene, something that doesn't rock the GPU, protag' got as far as meeting the Colonel and right before the big revelation , so about two/three minutes in, BAM! ... screen goes dead -- and by dead I mean it switches itself off citing a lack of signal -- and then, a few seconds later, I start seeing the Mobo hex-readout flickering its way back up to FF (it's booting). Huh. That's new. It reboots clean, I shut it down.
• Caturday, start PC, using Windoze apps, email, surfing interwebz. All is fine. Leave it sit for a couple of hours more, come back, still running okay.
• Start Tropico 4 demo, again, something unintensive. It's running at 700 fps. I play the tutorial. With nothing more than a static, graphical text screen and a Espanol-accented gentlemen explaining how he was taken under the wing by General A...- BAM! Oh, ffs.
• Reboots. Start browser. Start HWmonitor. Make comment to someone that "it's gotta be the GPU, it's never rebooted on anything other than graphics" while typing "+5 VCCH" into goog...- BAM! 'Ello, 'ello, 'ello, what's all this then?
• Why search for "+5 VCCH"? Because it says +5 but is only about 3.6v. Could that be a sign my PSU is failing as some have suggested? Hmm. Seems most forums don't trust HWmonitor's reading of voltages, so it might mean jack squat.
• PC restarting, but it only gets as far as the Windoze logo and stops. I wait 4 minutes more before the reset button is struck again.
• PC restarting, but AGAIN it only gets as far as the Windoze logo and stops. I wait an over-patient 20 minutes before the reset button is struck again.
• PC restarting, but I boot into Safe Mode. It's fine. And Event Log says nothing about being unable to start. There's no warnings other than the usual johci can't start, Windoze profile back-ups being used, and then twenty minutes of nothing important.
• Run Windows DVD, try to recover Windows partition but the blasted thing doesn't even recognise that my SSD, which it relegates to H drive, is an O/S-enabled drive. Fluff about with that, couple more reboots, and then shutdown.
• I take a strong torch and have a look in the back of the PSU. No obvious signs of fluff infestation. Fan appears to be a little dust-covered but not overly. No obvious signs of burnt components. *sigh*
• Start PC, enters Windoze fine. Och, I give in. Start typing this update.

tl;dr?

• PC is rebooting itself spontaneously AND/OR freezing
• Freezes came first, first starting weeks ago, and very intermittently. The screen would freeze and, depending (perhaps?) on the amount of sound caching, the sound might continue for a bit, or it might turn into a very loud digital staccato.
• Recently, I have also had half-freezes, in which the video and sound "scratch" in the hip-hop sense for a few seconds and then continue on. This happened thrice within the first ten minutes of playing a video game, and then never again, until hours later the game simply freezes.
• Reboots consist of a black screen, the screen that then switches itself off, citing a lack of video data, and then the Mobo rebooting.
• On earlier occasions (that is, until 24 hrs ago), the rebooting part would not happen before I reset 3-10 minutes later, and the black screen would happen a bit longer before the screen would switch-off.
• At first, thought it was GPU drivers (thus this thread's title), but the Event Viewer reports NO failures, only a short "The previous system shutdown at 6:11:24 AM on ‎14/‎07/‎2012 was unexpected." after a bunch of mundane information entries.
• Furthermore, the temperatures on the GPU and CPU and Mobo according to HWmonitor, fwiw, are all fine.
• This lack of event log information also precludes, to my mind, any other sort of software-related failure. This definitely appears to be "below" Windows's view, so hardware.
• SSD could also be a cause, though SSD might explain the video failures, but I am less inclined to think it would explain the consistent rebooting.
• The reboots do lead me to think PSU. Some voltage gets too low, too high, too wibbley-wobbley-timey-wimey and causes a failure first in the GPU and then a reboot by the Mobo. As the PSU has degraded, the time between GPU glitching because of it and the Mobo simply rebooting itself has grown shorter. Non?
• I can say for CERTAIN that it's not the memory: MemTest ran those suckers hard for 7 hours without a single glitch.
• Concluding speculations that it could still be a HW failure on GPU (mostly seems to happen in graphics), on SSD (graphics use the SSD a lot, too, y'know) or on PSU (it's rebooting itself).
• Only things not on the suspect list: Mobo, normal HDDs, case, pretty case/fan lights. Pretty sure the LEDs aren't the cause.

STILL tl;dr? *sigh* Fine. Sh!7's freezing and rebooting itself. The end.

It also occurs to me I have failed to give you the full spec's, though I'm not sure in this case they help:

CPU: Intel Core i7 950 3.0Ghz, 8MB Cache, 1066Mhz FSB, Quad Core
CPU cooler: Noctua NH-D14 CPU Cooler
Mobo: Gigabyte GA-X58A-UD5 MB, Intel X58+ICH10R, QPI 6.4GT/s, DDR3 2100, PCI-Ex16, SATA3, USB 3.0,RAID, IDE, 2xGbE LAN, 8ch audio, Heatpipe
Mem: Corsair 12GB Kit (3x4GB), PC-12800 (1600MHz) DDR3
SSD: OCZ 120GB SSD, 2.5", Vertex 2, SATAII
HDD: Samsung 2TB Spinpoint F4 Eco Green, SATAII, 5400rpm AND Western Digital 1TB, Caviar Black, SATA-III, 7200RPM
Case: CoolerMaster HAF X (so, 4 x 20mm fans or somesuch)
GPU: EVGA Geforce GTX580 (772Mhz), 1536MB GDDR5 (4008Mhz)...
PSU: Corsair CMPSU-1000HXAU 1000W ATX Modular Power Supply, 140mm fan, 3-Way SLI Certified, 6 PCI-E Connectors, Universal AC input 90~264V
UPS: Eaton 5110 1500VA USB 900W Line Interactive UPS

Any further thoughts or tests I could perform?
I have thought of pulling off the cover and testing voltages, but my multimeter is buried deep in a storage locker somewhere miles away.

This post has been edited by aragond on Saturday, Jul 14 2012, 15:09

[matthew1g]

back when I had my GTX480 It was rock solid and never glitched, even under extreme heat pressure (I don't have an AC in my room so room temperatures would fluctuate to around 30 degrees celcius).

The buzzing sound seems like an issue from the motherboard or psu to be honest. I know for sure that whenever i forgot to plug in the 8 pins from the psu to the GPU, the card would buzz loudly. The ATI 4870x2 I had before it would light up 4 LEDS it had on the board itself and would send a beeping noise through the mobo speaker.

With the ATI, I also had an issue where games would artifact, get BSODs, crash, stutter and generally get very low FPS. But running windows in general was fine. It turned out to be a bad ASUS motherboard. I switched that to gigabyte and I suddenly got perfect performance and much higher FPS in-games.

[Stinky12]

Remove your video card and check for dust. Remove the shroud if possible as dust tend to get stuck between the heatsink blades where using a dust blower don't really help that much.
Try your video card in then 2nd PCIe x16 slot on your motherboard and see if the problem still exist.
Run FurMark and check if the video card gives you any artifacts (color blocks or lines).
Try your video card on another computer (if you have a spare system to test it out).

For the SSD, try another SATA port or change the SATA cable.
Also disconnect that UPS and just run your system without and see if that helps. Maybe it's your UPS that's causing the problem?

This post has been edited by Stinky12 on Saturday, Jul 14 2012, 20:49

[leik oh em jeez!]

I love how you're so quick to bash nvidia. I'm running the 296.10 drivers with a GTX480 under Windows 7 x64 and haven't had any stability issues. I did however used to have a similar problem (though not nearly as bad) with my GTX275 COOP card. Resting an old CPU cooler on the backplate solved the problem. This was EVGA's design causing a problem with my specific install. I didn't go around bashing nvidia because of it. In fact, I sold the cards and bought another nvidia.

You'll want to keep an eye on temps when you run games, make sure you're getting optimal airflow to your GPU, clean the dust out of the heat sink, crank up the GPU fan, etc. and see if killing all background applications helps any. It's possible your card may be starting to die, or it could even be a f*ck up with Windows. If you have another GPU, test it. (That test could also be reversed, as in testing your GPU in another computer.) If you have an extra hard drive laying around, swap it in, install a fresh copy of Windows and see if the problem continues.

As far as your +5v reading, try a few different programs to see if they read it differently. Your system could be failing due to lack of enough power, which could be caused by a failing PSU or by bad circuitry in your house. PSU is an easy fix, and cheap depending on how much power you need.

This post has been edited by leik oh em jeez! on Saturday, Jul 14 2012, 21:46

[aragond]

Hey, all.
Yeah, sorry for bumping an old post, but it's only recently that this was brought to a head ... aaaaaand, then nothing was done about it.

QUOTE (leik oh em jeez! @ Sunday, Jul 15 2012, 07:43)

I love how you're so quick to bash nvidia. I'm running the 296.10 drivers with a GTX480 under Windows 7 x64 and haven't had any stability issues.

Y'know, that isn't helpful. What you're saying is if YOU'VE never had an issue with the same driver I'm using it couldn't POSSIBLY be the driver, because your computer is representative of aaaaaaaaall the world's computers, right?

Anyway, after all the crashes, all the crap, eventually, the machine started something new. I had started getting black screens, and that was followed by black screens as soon as anything remotely graphical happened, like, say, the Windows splash-screen. Funny thing is, you could still hear the disk flashing, and when I typed in my password, it would start the disk-up again, so it was logging in okay. And quickly this was followed by half-screens, where you get half the screen and that's all.

The last one was the clear indicator the gpu was toast. I rang my supplier and they agreed to pre-test it for RMA. Soon after, it was RMA-ed. So, there you go. Eight months of sh!t all because the GPU was slowly dying.

And then I waited. And waited. And waited.

And loooooooooooooong after submitting it, I get an email telling me... EVGA have tested the sh!t out of your GPU and they've found absolutely nothing wrong.

I said a few words about their "pre-testing" charade, gave them my computer and they tested the sh!t out of everything they could possibly find: the disks, the mobo, ran it hot for hours, updated some drivers, tweaked some settings, ran some more tests.

Nothing.

Now I have it back. It's not crashed the gpu yet, but I am waiting impatiently for the other shoe to drop. Frankly I think someone just voodoo-cursed it, some Nigerian "prince" I 415-bated or something, because little sh!t has happened since day 1 -- I still haven't (and nor have R* nor EA) been able to get LA Noire or MOH2 running on it. Just little weird sh!77y things.

Anyway. Just closing the loop.

[ccrogers15]

I had an nvidia card in a computer a few years back with 256MB video memory and it was PCI Express. The card would ALWAYS freeze and make the computer slow down. The solution? Connect the monitor to the OTHER DVI Port. Thats what helped me.

[yojo2]

Lol. Check out the second post in this topic :E

[aragond]

QUOTE (yojo2 @ Sunday, Dec 9 2012, 03:22)

Lol. Check out the second post in this topic :E

Yeah, apologies, I forgot to... or did I? I think my second post mentions it ... that I *did* try this and it DID work... for a while, and then kept degrading.