2x6970's Crashing Repeatedly with GUIMiner

[DeathAndTaxes]

Calm down. I opened up the case and the 100c card is getting 80c, and the 80c card is getting 70's-60's. It's just my case. One card runs great. But my second card is close to the bottom with minimal air flow.

Sounds good.  Like I said if temp drops it is air restriction.  You need a large volume of cool air to keep those GPUs cool.  If you need to use a case look for one which has plenty of room below the bottom card.  Also look for one which has the HDD mounted high (up near the CPU)  that gives you a straight shot from the front mounted fans to the cards air intake.

I've got my mining pretty much stable. I'm tweaking the voltages lower to see what is stable to get cooler temps maxed out.

Lowering memclock as far as possible (depends on the card, bios, company) will reduce some heat and give you some headroom.

This rig isn't going to be mining 24/7. It'll be mining overnight or when I'm in class. So it'll be running less than 24 hours per day at those temps and fan speeds. I have a 3 year warranty. Will that cover me if my fans malfunction? I would guess so. I'm not worried about cooking my GPU because anything above 90c is completely unacceptable to me. I try for 70'sc.

Well 16 hours a day is pretty much same as 24 hours a day.  Mining is hard on hardware.  Yeah a dead fan will be covered under warranty but it will be a pain in the ass and a long 2+ week delay getting your refurb card (which was used by someone else who likely abused it ). 

With good airflow you should be able to keep fans speed down to 75% or less.  That will improve longevity of the fan.  If your motherboard allows separating the cards more (because it has 3+ 16x slots) that helps too.  You can find extra long crossfire cables if you need one for gaming.

[1713883735]

1713883735

[1713883735]

1713883735

[Bananington]

Calm down. I opened up the case and the 100c card is getting 80c, and the 80c card is getting 70's-60's. It's just my case. One card runs great. But my second card is close to the bottom with minimal air flow.

Sounds good.  Like I said if temp drops it is air restriction.  You need a large volume of cool air to keep those GPUs cool.  If you need to use a case look for one which has plenty of room below the bottom card.  Also look for one which has the HDD mounted high (up near the CPU)  that gives you a straight shot from the front mounted fans to the cards air intake.

I've got my mining pretty much stable. I'm tweaking the voltages lower to see what is stable to get cooler temps maxed out.

Lowering memclock as far as possible (depends on the card, bios, company) will reduce some heat and give you some headroom.

This rig isn't going to be mining 24/7. It'll be mining overnight or when I'm in class. So it'll be running less than 24 hours per day at those temps and fan speeds. I have a 3 year warranty. Will that cover me if my fans malfunction? I would guess so. I'm not worried about cooking my GPU because anything above 90c is completely unacceptable to me. I try for 70'sc.

Well 16 hours a day is pretty much same as 24 hours a day.  Mining is hard on hardware.  Yeah a dead fan will be covered under warranty but it will be a pain in the ass and a long 2+ week delay getting your refurb card (which was used by someone else who likely abused it ). 

With good airflow you should be able to keep fans speed down to 75% or less.  That will improve longevity of the fan.  If your motherboard allows separating the cards more (because it has 3+ 16x slots) that helps too.  You can find extra long crossfire cables if you need one for gaming.

-I'm a gamer, this is a case built for gaming, plus I don't have the money for a new case just to mine.

-My memory clock is as low as possible.

-As for the RMA wait period, I'll always have one 6970 running to keep my games going. Lol

Thank you guys for your advice. I'll keep working on this stuff and update with what I finally get it stabilized at.

[Bananington]

Ok, so I'm tweaking the settings on one 6970. So when I run it, after about an hour or so, I get a long poll IO error reported by GUIMiner. So when I wake up to check on the mining, it has stopped mining several hours ago.

Any ideas? Normally with a crash, it doesn't show up as long poll IO error, right?

[Bananington]

So I've come to the conclusion that I can't run two cards plugged in at stock speeds. For some reason, GPU 2 can run at stock speeds, but GPU 1 crashes at stock speeds. The cards run perfectly at stock speeds when I only have one or the other plugged into the mobo. I've tested both PCI-e x16 slots individually and they both work and run each card at stock speeds without crashing, but when I have them both plugged in, GPU 1 CANNOT run at stock speeds for some reason.

Does anyone else think this is a driver issue? Because it doesn't seem to be a hardware issue.

[portron]

So I've come to the conclusion that I can't run two cards plugged in at stock speeds. For some reason, GPU 2 can run at stock speeds, but GPU 1 crashes at stock speeds. The cards run perfectly at stock speeds when I only have one or the other plugged into the mobo. I've tested both PCI-e x16 slots individually and they both work and run each card at stock speeds without crashing, but when I have them both plugged in, GPU 1 CANNOT run at stock speeds for some reason.

Does anyone else think this is a driver issue? Because it doesn't seem to be a hardware issue.

Need more info...
 
Mobo? Have you updated your BIOS or checked for issues with your particular model?

What OS / Driver version are you using?

Crossfire bridge installed?  Are you manging both cards with Afterburner and using identical settings?

What are your temps like now (since you had the issue before)?

[P4man]

SOunds like your powersupply isnt up to the task.

[jake262144]

What you say sounds about right, P4
If the PSU doesn't feel up to snuff, OP might be getting a lot of power fluctuations resulting in a crash.

Request you post PSU specs, including the manufacturer and model info.

[Bananington]

Mobo - http://www.newegg.com/Product/Product.aspx?Item=N82E16813131655
PSU - http://www.newegg.com/Product/Product.aspx?Item=N82E16817152045
OS - Windows 7 64bit
CCC Driver - 11.12

Crossfire Bridge is installed because I will eventually need to crossfire for my games, but crossfire is disabled via CCC.

My Temps are GPU 1(81c) and GPU 2(70c) this is with GPU 1's clock lowered from stock settings until it stops crashing.

My PSU has more power than is required, it shouldn't be a problem unless it is faulty. :O

Like I said earlier, both cards run at stock speeds perfectly on there own in either PCI-e slot with either pair of my 8-pin connectors.

I appreciate all of your responses. I hope we can figure this out.

[DeathAndTaxes]

81C is too high for 24/7 operation.  Are you downclocking the memory? 

[Bananington]

81C is too high for 24/7 operation.  Are you downclocking the memory?  

Everything is downclocked. Core voltage(without crashing), AUX voltage(without crashing), Memory Frequency(without crashing). I've already come to the conclusion that my case is at fault, causing the the extreme GPU temps because of poor airflow. I'm buying a new case ASAP.

But temperatures aren't the issue for me right now, why won't my GPU 1 run at stock settings when GPU 2 is installed?

My miner crashes GPU 1 when the clock is at stock speeds, my games crashes the GPU at core clock stock speeds when it's usage is above 80% as seen in Skyrim. The GPU runs perfectly at stock when I don't have another GPU plugged in. So what is the issue?

[DeathAndTaxes]

81C is too high for 24/7 operation.  Are you downclocking the memory? 

Everything is downclocked. Core voltage(without crashing), AUX voltage(without crashing), Memory Frequency(without crashing). I've already come to the conclusion that my case is at fault, causing the the extreme GPU temps because of poor airflow. I'm buying a new case ASAP.

But temperatures aren't the issue for me right now, why won't my GPU 1 run at stock settings when GPU 2 is installed?

Installed or mining.

GPU 1 crashes if GPU2 is idle or under full load?
What is the VRM temp (not core temp) on GPU1 when it crashes?

[Bananington]

81C is too high for 24/7 operation.  Are you downclocking the memory?  

Everything is downclocked. Core voltage(without crashing), AUX voltage(without crashing), Memory Frequency(without crashing). I've already come to the conclusion that my case is at fault, causing the the extreme GPU temps because of poor airflow. I'm buying a new case ASAP.

But temperatures aren't the issue for me right now, why won't my GPU 1 run at stock settings when GPU 2 is installed?

Installed or mining.

GPU 1 crashes if GPU2 is idle or under full load?
What is the VRM temp (not core temp) on GPU1 when it crashes?

-installed, and/or mining. Even if they are crossfired(I did this to test mining stability) GPU 1 still crashed.
-GPU 2 can be idle or under full load, it does not matter at all, just as long as it is plugged in the issues will occur. I have not tried switching the GPU's around. (putting GPU 1 in GPU 2's spot and visa versa) but I have tested them 1 card at a time in each slot and it worked fine for both at stock settings.
-How can I determine VRM temp of GPU 1? And what is VRM?

I am also using MSI Afterburner to manage/monitor these GPU's. I believe someone asked that earlier and I failed to answer. Crashing occurs when the cards have different and identical settings as well.

[worldinacoin]

I am also having problems with 6970, I stopped running 3 of them and instead left one for mining.  It is way too hot no matter how you undervoltage it etc.  I think the 5xxx series are better for mining.

[Bananington]

I am also having problems with 6970, I stopped running 3 of them and instead left one for mining.  It is way too hot no matter how you undervoltage it etc.  I think the 5xxx series are better for mining.

I think you misunderstand, my problem is not a temperature problem. My problem is that when I have two GPU's plugged in, one of them crashes at core clock speeds within seconds of usage going above 80% way before temps get out of the 70's.

[P4man]

Even though on paper your PSU is certainly powerful enough, it still sounds like a power delivery issue to me. Can you try another one?

[jake262144]

OP, read this article:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=246

The PSU reviewed there is build on the same platform as yours. They are both manufactured by Andyson according to the same spec. The only difference are stickers, just like with reference GPUs.

Although it might come as a shocker, contrary to what the label might imply, your PSU is NOT capable of delivering 1200W in continuous mode.

Moreover, your PSU has all its 12V power divided into four rails, each of them limited to 40A (480W).
You need to take very good care when loading up these rails.
If you connect your PSU in such a way that one rail becomes overloaded, as the PSU heats up and as a loses some efficiency (so-called de-rating), it might shut off on you.

This is a half-decent PSU but the rail distribution makes it quite hard to connect it the right way.
You'll need to read the manual and figure out which connectors to use to distribute the load evenly across the rails.

[Bananington]

Even though on paper your PSU is certainly powerful enough, it still sounds like a power delivery issue to me. Can you try another one?

No I can't.

OP, read this article:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=246

The PSU reviewed there is build on the same platform as yours. They are both manufactured by Andyson according to the same spec. The only difference are stickers, just like with reference GPUs.

Although it might come as a shocker, contrary to what the label might imply, your PSU is NOT capable of delivering 1200W in continuous mode.

Moreover, your PSU has all its 12V power divided into four rails, each of them limited to 40A (480W).
You need to take very good care when loading up these rails.
If you connect your PSU in such a way that one rail becomes overloaded, as the PSU heats up and as a loses some efficiency (so-called de-rating), it might shut off on you.

This is a half-decent PSU but the rail distribution makes it quite hard to connect it the right way.
You'll need to read the manual and figure out which connectors to use to distribute the load evenly across the rails.

I'm not familiar with PSU's at all. I have 1 video card plugged into the PCI-E slot on the PSU, and the other video card plugged into another PCI-E slot on the PSU. Wouldn't that distribute the load?

Any suggestions on how to test this would be appreciated.

[DeathAndTaxes]

Even though on paper your PSU is certainly powerful enough, it still sounds like a power delivery issue to me. Can you try another one?

No I can't.

OP, read this article:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=246

The PSU reviewed there is build on the same platform as yours. They are both manufactured by Andyson according to the same spec. The only difference are stickers, just like with reference GPUs.

Although it might come as a shocker, contrary to what the label might imply, your PSU is NOT capable of delivering 1200W in continuous mode.

Moreover, your PSU has all its 12V power divided into four rails, each of them limited to 40A (480W).
You need to take very good care when loading up these rails.
If you connect your PSU in such a way that one rail becomes overloaded, as the PSU heats up and as a loses some efficiency (so-called de-rating), it might shut off on you.

This is a half-decent PSU but the rail distribution makes it quite hard to connect it the right way.
You'll need to read the manual and figure out which connectors to use to distribute the load evenly across the rails.

I'm not familiar with PSU's at all. I have 1 video card plugged into the PCI-E slot on the PSU, and the other video card plugged into another PCI-E slot on the PSU. Wouldn't that distribute the load?

Any suggestions on how to test this would be appreciated.

It can be more complicated than that.

Your 6970 has 2x power connectors.  

Your PSU has 4 12V rails.  Each rail is only powerful enough to support 1 6970.  If you have both 6970s on the same rail and/or 6970 and mb on the same rail it will overload that rail and make the system unstable.

So you want it hooked up like this

Motherboard & other 12V loads - rail 1
First 6970 - rail 2
Second 6970 - rail 3

Figuring out which connector goes with what rail can be "tricky".  They often are poorly marked.  Looking at the wattage sticker on PSU, any documentation that came with it, and any specs provided online may help.

[Bananington]

Alright, I just switched the cards and they both run at stock speeds without crashing. I have no idea anymore. Problem solved so far. I'll post if I have any unforeseen issues later.

[Bananington]

Ok, so I switched slots, So GPU 1 is now GPU 2 and vice versa. So instead of the computer locking up when GPU 2 is running at 99% load at stock clock speeds, the video driver crashes and i get a notification saying that my driver crashed and was restarted. This is a step away from my computer crashing, now it's the driver. :/

Any ideas ladies and gentlemen?