2x6970's Crashing Repeatedly with GUIMiner

[P4man]

Im gonna ask again; are you running both GPU-Z and Afterburner? If you are, that would explain the extreme temperatures and crashing. ATI cards have a bug, when 2 apps are polling for (VRM only?) temps, it can cause vcore to spike up to over 1.6v. I know HD5000 cards are affected by this (I killed one myself because of this), I havent seen confirmation 6000 series are affected, but it wouldnt surprise me.

Why dont you try not running any of these monitoring apps, not even CCC and just use cgminer, and see what happens? Or guiminer if you must

[1715327274]

1715327274

[Bananington]

Im gonna ask again; are you running both GPU-Z and Afterburner? If you are, that would explain the extreme temperatures and crashing. ATI cards have a bug, when 2 apps are polling for (VRM only?) temps, it can cause vcore to spike up to over 1.6v. I know HD5000 cards are affected by this (I killed one myself because of this), I havent seen confirmation 6000 series are affected, but it wouldnt surprise me.

Why dont you try not running any of these monitoring apps, not even CCC and just use cgminer, and see what happens? Or guiminer if you must

I'm sorry for missing your question before. I've been running just MSI Afterburner.

So I reset everything to stock and closed MSI Afterburner. Then I ran the miner with GPU 2(formerly known as GPU 1 with the issues) and the display driver still crashed and restarted. I'm not sure how to not run CCC, I just left it in the system icon tray.

[P4man]

If you have switched the cards (and power cables), and its still the same card acting up, Id say its time for an RMA.

Im still clinging on to my hypothesis though, since you mentioned GPU-Z and afterburner, I suspect you killed the card by running 2 monitoring apps at some point, and the vcore shot through the roof. It took less than 30 minutes at 1.6v to kill my 5850. It didnt even die right away, for a week it kept running, but not very stable, far hotter than before and became ridiculously sensitive to voltage adjustments. Then it just died.

[Bananington]

If you have switched the cards (and power cables), and its still the same card acting up, Id say its time for an RMA.

Im still clinging on to my hypothesis though, since you mentioned GPU-Z and afterburner, I suspect you killed the card by running 2 monitoring apps at some point, and the vcore shot through the roof. It took less than 30 minutes at 1.6v to kill my 5850. It didnt even die right away, for a week it kept running, but not very stable, far hotter than before and became ridiculously sensitive to voltage adjustments. Then it just died.

I have not mentioned GPU-Z. I do not use GPU-Z. I only use MSI Afterburner to monitor my GPU's.

The card runs fine on it's own though. If I just have the one card plugged into either PCI-E slot it'll run at stock settings without crashing, It's when I plug the second GPU in, is when it acts up.

[P4man]

Sorry then, I must have misread or confused two threads. I would have sworn you mentioned gpuz.

I guess that puts us back to a cooling or power delivery problem. This might have been discussed before, but have your tried removing your case's side panel (if any) and pointing a big deskfan at it?  Just to narrow down the possible causes.

[jake262144]

I guess that puts us back to a cooling or power delivery problem...

...in which case:
OP, please try the following:
If you are using the hardwired PCIE cable, unplug it.
Use only the two modular cables to power the cards.

While you open up the case to re-cable the GPUs, do take care of proper cooling as P4man suggested.

[Bananington]

Sorry then, I must have misread or confused two threads. I would have sworn you mentioned gpuz.

I guess that puts us back to a cooling or power delivery problem. This might have been discussed before, but have your tried removing your case's side panel (if any) and pointing a big deskfan at it?  Just to narrow down the possible causes.

I have removed the side of the case and aimed a fan at it. This does not fix the issue.

I guess that puts us back to a cooling or power delivery problem...

...in which case:
OP, please try the following:
If you are using the hardwired PCIE cable, unplug it.
Use only the two modular cables to power the cards.

While you open up the case to re-cable the GPUs, do take care of proper cooling as P4man suggested.

I have done this. I have tried two seperate modular power cables, when this failed I have switched pack to 1 hardwired and 1 modular cable.




I find this really odd. Both PCI-E x16 slots on the mobo run fine, I have tried both 6970's in either slot running at stock speeds with the GUIMiner and Skyrim with no crashing, only 1 card plugged into the Mobo at a time though.

But when I plug two 6970's into my mobo, the same 1 crashes at stock clock speeds. If it is the primary VGA adapter, the computer completely freezes at stock settings when usage is above 80% and I have to hard restart. If the tricky card is the secondary VGA(the card my monitor is NOT plugged into), the video driver will crash and restart(without causing me to restart my computer) at stock settings when the usage is above 80%.

I have tried two separate modular power cords/ 1 modular and 1 hardwired power cable. My CCC driver's are up date with 11.12(I will be certain to update to 12.1 when they release it).


I am not completely against RMA'ing the card, but what if it isn't the card? Could it be the crossfire bridge, even though I have crossfire disabled?

[DeathAndTaxes]

You likely are looking at an RMA but to be sure I would do this:

1) I would remove the crossfire bridge.  Eliminate all other sources of failure.
2) Do a clean install of windows or try linuxcoin on usb drive.  Windows drivers don't like swapping cards around.  Don't ask me why.
3) When you do a clean install just install driver ONLY and APP SDK (Open CL).  No afterburner (bleh), no CCC.
4) Use cgminer w/ cards running at stock.

If you still get crashes and or hard locks shortly after mining then it is hardware but what hardware?

The confusing part is it could be:
a) MB
b) first card
c) second card
d) both cards
e) PSU

[ensign_lee]

I get that whole "display drive has crashed, but ahs recovered" problem with my 5870 that is slowly dying. I have to just keep starting it until it takes. I'm afraid to say it man, but I think one of your cards is just dying and that's why you're having these problems.

[Bananington]

You likely are looking at an RMA but to be sure I would do this:

1) I would remove the crossfire bridge.  Eliminate all other sources of failure.
2) Do a clean install of windows or try linuxcoin on usb drive.  Windows drivers don't like swapping cards around.  Don't ask me why.
3) When you do a clean install just install driver ONLY and APP SDK (Open CL).  No afterburner (bleh), no CCC.
4) Use cgminer w/ cards running at stock.

If you still get crashes and or hard locks shortly after mining then it is hardware but what hardware?

The confusing part is it could be:
a) MB
b) first card
c) second card
d) both cards
e) PSU

I'll try it without the bridge and post my results here.

I get that whole "display drive has crashed, but ahs recovered" problem with my 5870 that is slowly dying. I have to just keep starting it until it takes. I'm afraid to say it man, but I think one of your cards is just dying and that's why you're having these problems.

*sigh* I hope you're wrong. I don't wanna pay that damn shipping just out of principal. MSI's products should work like advertised.

[jake262144]

2) Do a clean install of windows or try linuxcoin on usb drive.  Windows drivers don't like swapping cards around.  Don't ask me why.

The fact of the matter is, windows uses hardware slot/port enumeration when ID-ing and configuring newly installed hardware.
Whether it's a GPU or a flash drive, changing the slot/port results in Windows discovering a new device. The drivers are already there so at least that is not a problem.

This approach looks goofy when you have 4 usb slots easily accessible and connect the same flash drive to the random port each time. Eventually, Windows will have quadrupled the device information.

However, this approach can be used for your advantage: should your device stop working correctly in one port due to some glitch that results in Windows "Not being able to find the correct driver" for a hitherto well-known device, you can always trivially reconnect it somewhere else and restore its functionality.
Of course, you could delve into the registry, find the faulty entry and delete it but such approach is not for the faint of heart.

3) When you do a clean install just install Linux

There, I fixed it for you  


OP, I'm not sure what information about the crash you are able to get out of the system, but do try to find out whether the fault seems to be card- or port-based.
That is, whether it is one specific GPU that crashes or any GPU in the specific mobo port that crashes.

[Bananington]

2) Do a clean install of windows or try linuxcoin on usb drive.  Windows drivers don't like swapping cards around.  Don't ask me why.

The fact of the matter is, windows uses hardware slot/port enumeration when ID-ing and configuring newly installed hardware.
Whether it's a GPU or a flash drive, changing the slot/port results in Windows discovering a new device. The drivers are already there so at least that is not a problem.

This approach looks goofy when you have 4 usb slots easily accessible and connect the same flash drive to the random port each time. Eventually, Windows will have quadrupled the device information.

However, this approach can be used for your advantage: should your device stop working correctly in one port due to some glitch that results in Windows "Not being able to find the correct driver" for a hitherto well-known device, you can always trivially reconnect it somewhere else and restore its functionality.
Of course, you could delve into the registry, find the faulty entry and delete it but such approach is not for the faint of heart.

3) When you do a clean install just install Linux

There, I fixed it for you  


OP, I'm not sure what information about the crash you are able to get out of the system, but do try to find out whether the fault seems to be card- or port-based.
That is, whether it is one specific GPU that crashes or any GPU in the specific mobo port that crashes.

It is a specific GPU that crashes.

[DeathAndTaxes]

It is a specific GPU that crashes.

Hmm.  I missed that.  So it is a specific card that crashes (always the same card that crashes) but it only crashes when two cards are installed?

Even if you don't want to use linux coin I recommend dropping it on a usb stick, booting from usb and trying.  If you get crashes there on the same card at stock you likely are looking at an RMA.  If you don't then windows is likely fubared and rather than waste a lot of time you probably should just do a clean install.

[Bananington]

It is a specific GPU that crashes.

Hmm.  I missed that.  So it is a specific card that crashes (always the same card that crashes) but it only crashes when two cards are installed?

Even if you don't want to use linux coin I recommend dropping it on a usb stick, booting from usb and trying.  If you get crashes there on the same card at stock you likely are looking at an RMA.  If you don't then windows is likely fubared and rather than waste a lot of time you probably should just do a clean install.

I'm going to try that, but last time I tried to run linux coin, I hung at the splash screen. No hardware has changed, I have reinstalled windows 7 since then. I'll try these things soon ladies and gentlemen. Thank you all for your troubleshooting tips.

[jake262144]

Hey, if LinuxCoin (or any other ready-to-go distro) can't get through the boot-up there's definitely something wrong with the hardware.

[Bananington]

I tried running the card without the bridge and I still have the driver crash and recover.

Linuxcoin hangs at the splash screen. I believe this may be due to a BIOS setting but I have no idea.

Anyone?

[P4man]

live linux distro's are fairly slow to boot, since the image is compressed; give it some time. You may also want to try another USB stick.
Last thing to do, when it seems frozen, try this keycombo
ctrl+alt+F1
Do you get a console? Or are your keyboard lights flashing?

[Bananington]

live linux distro's are fairly slow to boot, since the image is compressed; give it some time. You may also want to try another USB stick.
Last thing to do, when it seems frozen, try this keycombo
ctrl+alt+F1
Do you get a console? Or are your keyboard lights flashing?

I'm going to try and install it on another USB Flash Drive. So I'll probably be reporting back in 2 hours or so.

[Bananington]

Ok, so using a different flash drive worked. This is a flashdrive that I use exclusively for other things, any idea why the other flash drive didn't work?

I'll be testing mining out as soon as I figure out how to.

[Bananington]

GUIMiner won't connect when I try to mine. I've never mined with linux. Any tips?