[bumrush]

I've been having major issues with my 6870 which was operating extremely well for 3 weeks..

I'm at a complete loss and have exhausted pretty much every resource at my disposal (outside of testing the card in different machine) and need some help..

First of all, here are my system specs:

i7 2600K @ 4400 1.29v (tested with Prime95 blend for 3 days, no errors, fully stable, 36c / 72c under heavy load)

16GB Corsair Memory, Running at 1333, Memtest86 tested, no errors (4x4)

MSI 6870 920/1050, was overclocked to 950/1100 with no problems

Cooler Master GX-750 (I know this is not the best, didn't do enough research but it seems to have ample power)

Intel 510 SSD 120GB

Dual Samsung 1TB Drives, Raid 0

AsRock P67 Extreme 4

Problem:

I have been out PC gaming for many years and finally dove back in with the rig listed above.. For the first 3 weeks I played my fair share of COD Black OPS and had zero problems with the card.. I'm not certain if that specific game has a lower GPU load, but it ran stable. Around a week ago I started playing Portal 2 and noticed green dots on the screen.. I figured the OC was unstable so I went back to stock, which is 920/1050 on the MSI config.

Fast forward to today- Random screen splitting in 3d, with occasional problems in 2D, Furmark or Kombuster run perfectly fine overnight then crash within 3 minutes the next time I run them.. BF3 runs great then crashed after 5 minutes the next time I play. The crashes, split screen errors, red lines, red dots and black screens are random. I don't know if it's my PSU, the video, my MB..

I've tried the following:

Removed the card and placed it on the PCIe slot below it. Same problem.

Placed a different set of power cables in to the card.

Uninstalled and installed 11.9 / 11.8 / 11.10 beta using the standard removal tools and uninstallers that are commonly used.

Re-applied thermal paste which was a disaster.. Lowered core temps by roughly 8c under load.

GPUz shows 80-90 VDDC under heavy Furmark load, is that too high?

Downclocked card to 900/1050 same problem.

Installed MSI utility and increased voltage to 1.25 and kept lower clock settings.

+12v seems stable, dips to 11.88 under heavy load, which is within spec *according to AsRock tool*

The thing that is so strange is the random nature of how these crashes happen.. They seem to be related to higher loads, but I once I figure that out, the machine will start artifacting in windows before I launch any game.. Reboot and everything is OK - I can then game for 2 hours.. Reboot then problems come back..

However, as of the last 8 hours, the stability has declined. OCCT results in red dots and lines within 2 minutes, before the GPU really has time to heat up.. Yet upon reboot I can play Black OPS for an hour with no problems..

So what do you guys think? Is this a power supply issue or the card toast? One thing I want to mention is that I added a second CPU fan to Cooler Master 212+ to setup a push pull system right after the first sign of trouble.. In doing so I removed the MSI card.. Could I have shorted the card out when I took it out of the case???

EDIT: Fanspeed set to 45%.. I don't think this is a heat issue even though I have seen it go as high as 90c.. At first it crashed around 88 or higher but re-applying the thermal compound lowered high load temps, in addition to crashing under 70c

 

[Ken1649]

Quote:

Originally Posted by bumrush;15255592
I've been having major issues with my 6870 which was operating extremely well for 3 weeks..

I'm at a complete loss and have exhausted pretty much every resource at my disposal (outside of testing the card in different machine) and need some help..

First of all, here are my system specs:

i7 2600K @ 4400 1.29v (tested with Prime95 blend for 3 days, no errors, fully stable, 36c / 72c under heavy load)

16GB Corsair Memory, Running at 1333, Memtest86 tested, no errors (4x4)

MSI 6870 920/1050, was overclocked to 950/1100 with no problems

Cooler Master GX-750 (I know this is not the best, didn't do enough research but it seems to have ample power)

Intel 510 SSD 120GB

Dual Samsung 1TB Drives, Raid 0

AsRock P67 Extreme 4

Problem:

I have been out PC gaming for many years and finally dove back in with the rig listed above.. For the first 3 weeks I played my fair share of COD Black OPS and had zero problems with the card.. I'm not certain if that specific game has a lower GPU load, but it ran stable. Around a week ago I started playing Portal 2 and noticed green dots on the screen.. I figured the OC was unstable so I went back to stock, which is 920/1050 on the MSI config.

Fast forward to today- Random screen splitting in 3d, with occasional problems in 2D, Furmark or Kombuster run perfectly fine overnight then crash within 3 minutes the next time I run them.. BF3 runs great then crashed after 5 minutes the next time I play. The crashes, split screen errors, red lines, red dots and black screens are random. I don't know if it's my PSU, the video, my MB..

I've tried the following:

Removed the card and placed it on the PCIe slot below it. Same problem.

Placed a different set of power cables in to the card.

Uninstalled and installed 11.9 / 11.8 / 11.10 beta using the standard removal tools and uninstallers that are commonly used.

Re-applied thermal paste which was a disaster.. Lowered core temps by roughly 8c under load.

GPUz shows 80-90 VDDC under heavy Furmark load, is that too high?

Downclocked card to 900/1050 same problem.

Installed MSI utility and increased voltage to 1.25 and kept lower clock settings.

+12v seems stable, dips to 11.88 under heavy load, which is within spec *according to AsRock tool*

The thing that is so strange is the random nature of how these crashes happen.. They seem to be related to higher loads, but I once I figure that out, the machine will start artifacting in windows before I launch any game.. Reboot and everything is OK - I can then game for 2 hours.. Reboot then problems come back..

However, as of the last 8 hours, the stability has declined. OCCT results in red dots and lines within 2 minutes, before the GPU really has time to heat up.. Yet upon reboot I can play Black OPS for an hour with no problems..

So what do you guys think? Is this a power supply issue or the card toast? One thing I want to mention is that I added a second CPU fan to Cooler Master 212+ to setup a push pull system right after the first sign of trouble.. In doing so I removed the MSI card.. Could I have shorted the card out when I took it out of the case???

EDIT: Fanspeed set to 45%.. I don't think this is a heat issue even though I have seen it go as high as 90c.. At first it crashed around 88 or higher but re-applying the thermal compound lowered high load temps, in addition to crashing under 70c

You have tested everything and I don't think the issue is related to driver but more on the power draw and temperatures (GPU core, VRAM and VRM) and stability.

If you have the option Power Setting in CCC Overdrive, try to increase the power set limit to +20% and see how it goes. The factory seems a bit high on the core, back it down to 800MHz.

 

[bumrush]

Thanks Ken... Not sure what you mean by power settings, my CCC doesn't have that option.. Also, reference 6870 clock is 900, I would rather purchase a new card then downclock it to 800 mhz, especially with BF3 around the corner.

One other thing to note.. I walked out of the room when doing a Kombuster stress test and when I came back the screen was red with wavy lines and dots.. Upon reboot the issue persisted, and a few minutes later I had no signal at all. This morning the PC booted up fine and I played an hour of Portal 2 with no problems.. I did find that if I ran the stress test for a longer period of time after the initial artifacts, that the problem did persist for several minutes after reboot.. If I closed or rebooted the computer quickly once the problems appeared, Windows seemed to be fine upon reboot.

Does this indicate that the PS is OK since the issue persisted after a hard reset / multiple reboots under a light load?

 

[Ken1649]

The Power Setting is in CCC Overdrive if you have it enabled (not sure if this option is available for a 6870). I think it is power and temperature related because it happens only at heavy load and the GPU pegged.

VRAM or VRM temps as the Core doesn't seem to be the problem. Take it out and test it in another system will save you lots of time from trouble shooting. There is a possibility the card is going bad though.

The reason to downclock is to narrow it down to the possible cause.

 

[bumrush]

Thanks Ken.. No power option in CCC.. I also tried MSI Afterburner, no option either.. Is there a different program you recommend?

 

[Ken1649]

Can you see this in CCC? You have to enable OVerdrive.

 

[bumrush]

No. Not visible... I'm assuming this setting allows for more juice to be pumped to the card, correct?

 

[Ken1649]

Yep.

You can try this, though I don't think it is related.

Try to clean the driver and registry with this:

1. Remove all CCC driver. Restart PC.

2. Clean it with Atiman Uninstaller v.6.0.0.msi. Download LINK

Run it with administrator privilege. Some antivirus software will give a false positive. Let it restarts to complete uninstallation.

3. Change PCIe slot.

4. Install Catalyst 11.9.

If you are installing 11.10 preview drivers, set the UAC to minimum.

 

[bumrush]

Done..

Same problem.. 30 seconds in to OCCT tool, red screen with wavy lines and dots..

Temps never even had time to hit 50c.

VVDC Current hit 70 in GPU-z...

Downclocked to 850..

Bad card or power supply? The OCCT tool really stresses my power supply.. +12V seems stable, at least according to the supplied AsRock software..

 

[Ken1649]

See if you can put a fan to hit directly at the VRAM and Vortage Regulator and still the same issue, the card is going bad. Sorry. Overvolting, core/memory downclock didn't solve it. Doubt it is the PSU. Final test is with different PSU or take the card out test it in another system.

 

[bumrush]

Thanks for all your help...

Last thing.. I just put my case side panel back on while and nudged the PC to the side.. Guess what? Same red lines with very slight touch to the case.. The card is attached properly, screws tightened and firmly in place in the PCIe slot.. Come to think of it, the same thing happened when I tested it in a different PCIe slot.. Could the motherboard have some sort of short?? I know I'm gasping for straws, but I don't want to rule out anything before I RMA the card.. Would hate go through all the trouble with MSI only to find out that the replacement card is having the same issues.

 

[Ken1649]

The only way to answer that is to test it in another system. You are taking the right steps to rule out everything else before RMA.

 

[bumrush]

Woah.. Now entering the twilight zone..

Removed the second screw from the lower part of the card which looked to be slightly bent.. Tightened the top screw as much as possible to prevent the card from sagging down... Just ran 10 minutes of OCCT with no crashing..

Could there be a short in the card or the MB/case? Something not grounded properly?

 

[Ken1649]

Sounds like a loose contact. Play around with it a bit will find out for sure. Or the PCI bracket is over stressed or PSU ground cable to the wall. Check everything.

 

[bumrush]

Yup.. It has to be... I closed the case and then ran OCCT and boom, immediate red screen and flashing red..

What is the correct method to clean the contacts on the actual card? As stated, the problem also occurred on the 2nd PCIe slot..

 

[bumrush]

Update: Card is now dead.. Shipping it to MSI for repair.

 

[Ken1649]

Sorry to hear that, hope they fix it for you soon.

 

[xPrestonn]

Quote:

Originally Posted by bumrush;15286084
Update: Card is now dead.. Shipping it to MSI for repair.

ouch. GL with the rma, sounds like you just got unlucky.