Possible Bad GPU? (Video/audio cuts out, computer still running)

I'm having what I believe to be a hardware issue, but I can't be certain.  I keep experiencing crashes, which cut off the signal to the monitor, but the computer keeps running, though it seems to be idling. 

The last five or so crashes all happened today.  I can all but guarantee that heat is not a factor.  My hardware monitors were reporting nice and balmy temps on everything right before crashes.  The last few crashes happened with the case open, and after the crash I checked myself, and even the typically hottest components were cool.

For the actual "crash", I'll be using the computer and it'll just cut video.  No BSOD, no reboot, no audio glitches, and no evidence of any attempts at input having any effect.  The monitor stops receiving input and goes black.  The computer is still running, though the fans slow a little (likely just my video card fan slowing since I use Sapphire Trixx to adjust the fan speed to run a little above default to keep temps down).  The last crash, I was watching a video, and the audio cut out a few seconds (2-5) after the video did.  The fact that audio cut out as well made me think it's a bigger problem than just my monitor potentially going bad.

On the note of my monitor, I have had another issue with it which makes me think it's on its way out.  While I don't think it's the cuprit, I'll describe that issue just to be thorough.  Sometimes when waking from a sleep state, it'll display static.  Just a multicolored mess of pixel gibberish.  It typically starts at the top and fills the screen downward like following a scan line.  The fix to this, has been simply power cycling the monitor.  Turn it off for a few seconds, get it to reboot and It'll work fine from then on out.  The fact that power cycling the monitor works, makes me believe that this issue is isolated to the monitor itself.

After getting the crash, I have to reboot with a hard reset.  I've had it crash again while in the process of rebooting.  Once, if I recall correctly, it crashed before reaching the desktop, and every other time it's crashed while rebooting, it's been after reaching the desktop but before all background processes have finished loading.  Most crashes have been a random time during use.

Right now I'm typing this up in safe mode, and haven't crashed yet thankfully.  But not sure that will last.

I built this rig in 2011, and have had little to no issues 'til now.

System Specs:

MOBO - ASUS Crosshair IV Formula

CPU - AMD Phenom II x6 1100T Black Edition

GPU - Sapphire Radeon HD 6970

RAM - G.SKILL Ripjaws Series 8GB (2x4GB) DDR3 SDRAM DDR3 1600

PSU - Corsair TX Series 750W ATX12V

System Drive (C:) - Western Digital VelociRaptor 150GB


I recently updated to the latest non-beta AMD driver for the video card: AMD Catalyst Omega 14.12 win7 64bit.  I'll be installing the latest beta (14.11.2) after posting this, to see if it makes a difference.

I'm not sure where to go from here in figuring out what exactly the problem is and would greatly appreciate any help that anyone can provide.

EDIT:

Changing to the 14.11.2 beta driver didn't help.  Going to try an older one, but at this point doubt it'll do any good.

EDIT 2:

Haven't experienced a crash in safemode yet.  One thing that just occurred to me, is that when the screen goes black, it's almost exactly the same as when the screen goes black when installing a new driver.  Further reinforces that it may be GPU related, but not sure if it's a hardware or software problem.

Hi toyotomy, Sorry to hear of the problems with your computer.

The fact that your computer seems to be Ok in safe mode is diagnostically significant…..

So give it a thorough extended test in safe mode to see if the improvement is enduring…. However please be very careful. especially if using safe mode with networking as your protections will not be fully operative and so you will be more vulnerable to infection.

Use the machine only for controlled testing until you have results that you can rely on.

Post back and let us know how you are getting on and we will advise the next steps.

Regards

paws

Please also take a look in your event viewer.  Your description can send a person to multiple conclusions.  Is it software or hardware?  If its a driver, then that should be indicated in the System event log when the event occurs.  You should review the logs for what happens around the time of the crash.

Here's how to open Event Viewer: http://windows.microsoft.com/en-us/windows/open-event-viewer#1TC=windows-7

While AMD drivers can be somewhat problematic, I'd suspect the video card is going bad or to a much lesser extent the power supply.  Try reseating the video card (remove it, reinsert it a couple of times).  Do the same for your RAM modules.  Just double check all your other connections.  It doesn't hurt to remove and replug things.

Safe mode alone is not enough to eliminate hardware problems because the hardware is utilized in a much different way in safe mode thus it could "mask" a hardware problem.

Last night, I uninstalled all video drivers then ran DDU in safe mode to clean up the leftovers.  I then installed 14.9 drivers (which worked fine for me in the past).  After a reboot, I got another black screen within 20 minutes.  I then gave up for the night.

Today, I booted the computer up to review the Event Viewer and test it in safe mode.  My monitor was giving me issues, so by the time I got it working it was too late to force safe mode so I just looked at the event viewer.  I also figured, that since I never took note of exactly when a crash occurred before, I'd let it run until a crash which by now were pretty reliable, so then I'd have something more specific to look for in the viewer.  Well, it didn't crash, though I found one minidump file, but it was corrupted, and WhoCrashed couldn't read it.  The event viewer did show the bugcheck:

0x00000117 (0xfffffa8007d5b450, 0xfffff88003d5d038, 0x0000000000000000, 0x0000000000000000)

Didn't give me a whole lot to go on other than what I already knew.  Sifting further through the event viewer, there were plenty of errors (one related to a program that's no longer installed but should nonetheless be innocuous), but one caught my eye.  Radeon Pro, a program to basically replace the Catalyst Control Center, I was using it to make Assassin's Creed Unity run decently while not looking like utter garbage.  It showed multiple errors.. so I uninstalled it, and rebooted.

I've been watching videos and whatnot for a while now and so far, no crash.  Without further testing, I'm not entirely convinced that the problem is solved though.  I wonder if it was some sort of issue between changes made by the new Omega driver and the seriously outdated Radeon Pro software.  If I get no crashes over the next day or two I might try re-updating to the Omega driver and see what happens.

EDIT:

Still a little strangeness.  Opened 7 or so wikipedia tabs in Firefox, and all of a sudden it changed color mode on me and the GPU was going like a bat outta hell.  ~82% usage for basically no good reason.  It didn't get hot.  Fan sped up and kept it cool… but still strange to have that happen.

I've had three crashes today thus far.  Woke up to the system unresponsive.  Rebooted on my way out of the house, and came home to it unresponsive.  And finally, I just rebooted it and it crashed right after it finished booting up (just opened a firefox window to check this thread and it crashed before I could get that far).

No minidumps.  Here are the warnings I'm getting in the Event Viewer:

The server service was unable to recreate the share My Apps because the directory D:\BlueStacks\UserData\Library\My Apps no longer exists.  Please run "net share My Apps /delete" to delete the share, or recreate the directory D:\BlueStacks\UserData\Library\My Apps.

The SBSD Security Center Service service failed to start due to the following error:
The service did not respond to the start or control request in a timely fashion.

A timeout was reached (30000 milliseconds) while waiting for the SBSD Security Center Service service to connect.

The AODDriver4.3 service failed to start due to the following error:
The system cannot find the file specified.

The AODDriver4.2.0 service failed to start due to the following error:
The system cannot find the file specified.

Custom dynamic link libraries are being loaded for every application. The system administrator should review the list of libraries to ensure they are related to trusted applications.

Name resolution for the name projects.scottsplayground.com timed out after none of the configured DNS servers responded.

__________

I have no idea what that scottsplayground thing is.

The Bluestacks program was uninstalled long ago.  I'm not sure how to make that event stop occuring.

Not sure what is causing the black screens.  I've checked cable connections, and they're fine.  I haven't tried removing and reseating the card yet.

Hi, you resolve the bluestack message (which is harmless) by doing what it says:

 

 

Please run "net share My Apps /delete" to delete the share….

 

Run this command from an administrative command prompt and it will be gone.

Next, I don't know what else to say.  The spybot service is hanging up.  Can't see that causing this problem other than a performance issue.  Next, you have an AMD Overdrive driver not loading.  This could be a driver issue, but I'm totally leaning towards a bad video card.

You need to go ahead and clean and reseat everything in the system and see if it helps.  Otherwise, I think you're due for a video card.  Or a warranty service if you're lucky enough to still be under warranty.  I experienced similar behavior with bad RAM though in the past.  You could download and run MemTest86.  If you're technical enough you need to create a bootable disc and boot the disc and run the test for several hours.  But, without a doubt you need to reseat all that stuff and make sure it isn't simply a bad connection.

I tried the "net share My Apps /delete" line.  It gave me "This shared resource does not exist." and More help is available by typing NET HELPMSG 2310" in response.  When I used "NET HELPMSG 2310", it again gave me "This shared resource does not exist."

I removed and reseated the video card, and got rid of what dust was around the PCIe slot.  Just booted up from doing that a few minutes ago.  I got the card in 2011 so I think I'm outta luck for a warranty.  I have a MemTest86 CD floating around somewhere already (did fairly thorough testing on all my components when I built this rig), so I'll run a nice long test with that soon.

Since it's probable that I'll need a new card… Anyone got any recommendations for a budget card that'll outperform a Sapphire Radeon HD 6970?  Must be below $200, under $150 preferrably.  Also, can't physically be any longer than that 6970… the one I have just barely fits in my case.  I'm starting to look into it myself, but I haven't been paying attention much to what's what in the current offerings before just now.  Was hoping to wait at least a few months or so and maybe build a new rig, but I'll need something to last me 'til then.

Put the "My Apps" in quotes just like I did.

A thorough test at the time you built the machine does not indicate the memory is still good.  Memory does fail, however rare it may be.

Reseating the RAM is also necessary - not just the video card.

You can see a thread about testing memory here that might help: http://forums.whatthetech.com/index.php?showtopic=103823

The tutorial uses Windows Memory Diag, but I recommend MemTest86 now that machines are mostly over 4GB of RAM now.

A lengthy test (over 12 hours) with MemTest86+ returned no errors.  Unlikely RAM is an issue.

Looking more definite that it's  a card going bad.

net share "My Apps" /delete  returned the messages "System error 3 has occurred." followed by "The system cannot find the path specified."

Searching for a replacement video card.

I bought and have since installed a Sapphire R9 270x 4GB.  Installed on New Years Eve, It's been running continuously since.  Initially it ran perfectly, I've had no crashes, black screens, or anything of the like.  However there's one thing that's been happening that I just noticed today and it has me a little worried honestly.

Every now and then the screen flickers for a split second.  Just a single frame of, from what little I can tell, is predominantly-white static.  It could just be the monitor, which I know is already on its way out (the previously mentioned problem of it displaying static on waking or being powered on, which I've had happen when it's not even connected to the PC).  But I'm nonetheless concerned about it being something else.  Is there a good stress test utility for GPUs or something I could use to suss out if there's a problem with the system besides just a slowly degrading monitor?

EDIT:

When the flicker/blips happen (I've just been reading reddit in Firefox while all this is happening), the window shifts a little.  Like it's been very quickly resized… it's really odd and kinda hard to describe.

EDIT 2:

I just remembered FurMark.. I'll use that a bit and see what happens.

EDIT 3:

Good scores in FurMark.  Didn't notice the flicker at all while running the tests oddly enough.

Until appleoddity comes back, I reckon you would be well advised to swap out your existing VDU for a known good working model and test it for a while.

If you don't have a spare known good working VDU handy, then maybe you could ask a chum to lend your theirs for testing purposes?

Regards

paws

I don't have another monitor to plug in at the moment, but I may be able to get one in the next few days.  In the meantime, some more details of what I'm seeing.

The "glitches" I'm seeing seem to be most frequent when using Firefox for whatever reason.  I've played a couple of video games and noticed no flickering or anything during those… which is odd.  Just a few minutes ago, after getting 2-3 flickers in about a minute, Windows turned off window transparency on me.  Nothing was using an ungodly amount of ram (most was Firefox, which was ~300MB) at the time.  It didn't give the usual pop up asking whether I want to change options like its done in the past when I've seen this happen (only rarely, usually with a poorly optimized game that's in the process of crashing).  It took a few minutes before it went back to normal… but I'm still not sure what prompted it.

I've had the screen flicker a few times while typing up this post.

I just did a search and found the suggestion to use Sapphire Trixx to disable ULPS (which would make sense if that's the culprit, since the GPU is under very low load when the flickering happens, and haven't seen flickering while playing games).  I checked that option, and see how it goes.

EDIT:

Still flickering after clicking the option to disable ULPS… and also not sure it's actually disabling ULPS.  In fact, I'm not sure TRIXX is doing jack squat other than adjusting fan speeds despite being supplied straight from Sapphire (yes it's the latest version).

Sitting here watching the GPU Core Clock change from moment to moment.  The "Force constant Voltage" option might have worked, but it's currently stuck at 0.950V… while the settings in TRIXX says 1225 for VDDC.  I can't find a good source for the stock voltage of this card, but what little I've found says 1.2v is stock for the 270x.  So it seems my card is undervolted?  Reloading and reapplying the default profile (the one that reads VDDC 1225), doesn't change the voltage.  Still at 0.950V.

Watching GPU-Z, I've noticed the flickers seem to correspond to the GPU load settling under 10% for a few moments, and then spiking above 20-30% or so.

EDIT 2:

Guess it's a 2D vs 3D settings thing with the different voltages and clock speeds.  Start up a 3D program and everything jumps up where it should be.

Going off of that for some further searching, it seems that this is a common problem for the R- series cards… Wish I'd delved that deep into researching them before purchasing.  THIS post seems to indicate that it could be a problem with the vBIOS supplied on the cards.  But there doesn't seem to be a straight forward way to get a working vBIOS… (never messed with vBIOS before either, and it seems a little risky since I don't even have on-board graphics or a different working card to fall back on should something go wrong.)  There doesn't seem to be any clear cut solution to this flickering issue if there is one at all.

If I can't find a way to fix it, hopefully Newegg will accept an RMA request, and I can then get an Nvidia card… as much as I dislike Nvidia.

Yes, all understood,

If the problem still exists with a known good working VDU and its cable! ( don't forget that cable and its connectors…)

Then trying another video card would be a wise move.

I'm assuming here that you have not overclocked… is that right?

A further and "no cost" step would be to boot from a Live distro of Linux…. Ubuntu may be a good choice, but use the distro you are most familiar with…

You will be able to carry out a pretty thorough test ( fingers crossed or knock on wood) using a live distro.and you should be able to see if the problems remains…. if it doesn't then its pointing to Windows or your drivers/services as a likely culprit.

however if the problems continues then its likely hardware as you suspect.

regards

paws

You may be dealing with multiple problems which is making it difficult to troubleshoot.

The brief transparency mode problem is indicative of a driver crash and recovery.  This would've caused a blue screen in XP, but in 7 it can recover.  However, you are usually given a prompt in the lower right hand corner saying your display driver crashed or stopped responding.   The event is also logged in the system event log.

The screen flickering issue is strange.  You really need to hook up a different monitor.  This is too similar to the other problems you know your monitor already has.  You need to be diligent to eliminate problems you already know for sure exist.  This may make troubleshooting the rest of this trouble easier.

I'm not sure which way to go with this.  I do think you are having a hardware issue, but I'm not convinced it is the video card.  These problems can be difficult to troubleshoot and often times require swapping parts and running for a while to see if it is fixed.  I can tell you that 12 hours of memory tests, or even 12 days of good memory tests means absolutely nothing.  I've had memory pass tests like this and still be bad.  In fact, the system I am currently typing on used to have a video driver (transparency issue) crash really often.  I tried absolutely everything, until one day I said enough is enough, took the machine to my shop and figured it out.  It was bad RAM.   And the only way I found that out was by swapping the RAM and running it for a while.  You could do something similar by running on one of your two RAM chips, then swapping them to see if one or the other causes trouble.

I'm concerned about talking about voltages and GPU speeds at all.  My rule is that when somebody brings a computer to me for troubleshooting something like this everything gets reset to defaults and all special tweaks get uninstalled and turned off.  I know its fancy to run all this stuff, but its usually problematic in one way or another.  I wonder if any additional items have been tweaked in your BIOS?  Everything needs to be reset and configured as defaults for me to be comfortable with it.  I'm good with BIOSes so I know how to reset them to default and adjust the important settings where they need to be.  Sometimes that includes RAM timing, etc.  But, your motherboard should be compatible with that RAM and should time it properly on its own.

These are all ideas, but I'm just pointing out this can be a slippery slope of hardware swaps and extended testing.  That's just the way it is.  Its tough for an end user that doesn't have these extra parts just sitting around.

Your idea of going to nVidia is great.  Not because I prefer nVidia over AMD (funny how we all have our opinions) but the fact that you would be eliminating AMD software all together.  I personally have had a lot of trouble with AMD drivers.  If it is a driver issue in anyway, this should eliminate it.

I haven't overclocked the card at all.  The only thing close I did was to set the "force constant voltage" settings… which didn't do anything anyway, and have since toggled it back off.  The video card is as it came stock out of the box (though technically the card comes overclocked).  The only thing I've modified at all, is adjusted the fan speed curve in TRIXX as it was running just a little warmer than I'd personally like.

What you said about resetting BIOS (Version 2002 which was what it shipped with) actually made me realize something.  I'm fairly certain my BIOS is running in default settings.  When my computer shut down unexpectedly multiple times (can't remember what made it happen the first time… think it was a game, most recently it happened with the old video card.), it would occasionally tell me to either reset BIOS to default or to enter setup and adjust it… I just selected to enter setup, then saved and exited.  But I just opened up some hardware monitors and my DRAM frequency is 668.9MHz, and Bus speed is 200.63(ish)MHz according to CPU-Z.  My RAM should be showing up as 800MHz (1600).  When I set the computer up, I had some instability issues, which I eventually nuked out (though for the life of me I can't remember what resolved it).  I eventually ended up bumping up the CPU Bus for the sake of the ram, then dialed back the CPU clock back to default.  But for the moment anyway it seems everything is default.  I'll have to review my old settings and compare what's set now, but it definitely isn't what it's meant to be.

After realizing that about my BIOS likely running at default settings, I'm wondering if I accidentally uninstalled my chipset drivers when I uninstalled my AMD graphics drivers.  Any way to see what version is currently installed for the chipset drivers?

I looked up the newer series of nvidia cards, and it seems even some of them are encountering this 2D flicker.

To be honest, I was just slacking off the last couple of days, so haven't done much troubleshooting… but I did do one thing.  I was playing a game that only uses 2D graphics (Don't Starve), so the card never went into 3D mode.  I was getting the flickering problems during that.  It's the first time I've seen the flicker in any game.  So I wanted to experiment and force it into 3D without actually tweaking the card… just giving it something 3D to render.  I started up Furmark, and ran it at the lowest resolution and settings possible.  Since it's a 3D test, the cards voltage and clock speeds went up tho the full 3D specs.  No more flicker in the 2D game with Furmark running minimized.  Even typing this out in Firefox (where I usually see frequent flickering), there's no flicker with it minimized.

I read some posts on the Sapphire and AMD forums and while there's no clear cut fix overall, the consensus seems to be that the problem really does lie in the card downclocking for 2D rendering.  Some suggest contacting the manufacturer for a vBIOS to flash (some of which have resulted in dead cards, which I don't have a secondary card to correct such a mistake, and messing with the BIOS voids the warranty), while others have advocated  changing the 2D clock to match the 3D clock in the CCC Profile.  Some have said the MSI Afterburner has a setting to force the card to run at 3D settings at all times (haven't looked into that one much… never use that program before.), and while looking up 2D flicker for nvidia cards, they recommended using EVGA PrecisionX (which doesn't work with AMD cards anyway) to force the card to stick with 3D stats.

I'm not discounting anything yet, but figured I'd update with what little I've tried and read.  I'll do more over the next couple of days.