CGminer, "dead" gpus

[bobbobobbo]

I feel this is better in this section. I have 3 Sapphire 7950s, along with a Gigabyte ga-z77x-ud4h board running on a 1050 gold rated psu.

Everything is stock clocked, I am using CGminer. I can only seem to run at around 15 intensity... (300k/hash). When I increase it to 20 intensity (520k/hash), I slowly get "GPU sick" then eventually "GPU dead" on all 3. Running at around 65% fan speed, the cards are all kept under 75C, even at 20 intensity.

Any help is greatly appreciated!!

[1714153924]

1714153924

[1714153924]

1714153924

[os2sam]

I feel this is better in this section.

What do you mean by that?  Better in the "Mining software (miners)" section instead of the CGMiner section?

Why?

[ssateneth]

When I increase it to 20 intensity (520k/hash), I slowly get "GPU sick"

So don't do that. problem solved.

[Wilderness]

I just bought 2 7950s and when mining in CGminer one of them goes sick / dead

I was running a BAT file in windows that works for a 4 x 7970 rig no probs and a 68xx, however sick for the 7950s. I just tried mining in Guiminer scrypt and the second card looks like its doing the same.

I have the latest amd drivers and just installed latest sdk not sure what else to try. Might try swap the cards to the board that I have 7970s with mining fine and see if the board is a factor

[Wilderness]

ok I switched the 2x 7950 with the 2x 7970 and the SICK followed the 7950 but the 7970s were fine in the other board.

is it worth trying old drivers and old SDK?

[os2sam]

ok I switched the 2x 7950 with the 2x 7970 and the SICK followed the 7950 but the 7970s were fine in the other board.

is it worth trying old drivers and old SDK?

I sounds to me like you got a dead GPU.

But it may be worth while to try older catalyst and SDK as I think there was an issue with the latest.  Did you check the readme file or post in the CGMiner thread?

[Wilderness]

ok I switched the 2x 7950 with the 2x 7970 and the SICK followed the 7950 but the 7970s were fine in the other board.

is it worth trying old drivers and old SDK?

I sounds to me like you got a dead GPU.

But it may be worth while to try older catalyst and SDK as I think there was an issue with the latest.  Did you check the readme file or post in the CGMiner thread?

Hi os2sam,

thanks for your reply, I tried my cards individually on a new mobo / install and one of them still gets sick. The other seems to be fine, has been mining for a while without failing.
I haven't tried the sick card for gaming but it appears to work fine until mining when it gets the AMD driver has stopped responding. Could you RMA for this reason? I thought they just test it and say well it posts therefore it's not faulty?

[os2sam]

ok I switched the 2x 7950 with the 2x 7970 and the SICK followed the 7950 but the 7970s were fine in the other board.

is it worth trying old drivers and old SDK?

I sounds to me like you got a dead GPU.

But it may be worth while to try older catalyst and SDK as I think there was an issue with the latest.  Did you check the readme file or post in the CGMiner thread?

Hi os2sam,

thanks for your reply, I tried my cards individually on a new mobo / install and one of them still gets sick. The other seems to be fine, has been mining for a while without failing.
I haven't tried the sick card for gaming but it appears to work fine until mining when it gets the AMD driver has stopped responding. Could you RMA for this reason? I thought they just test it and say well it posts therefore it's not faulty?

I've never had to RMA a GPU yet so I don't know.  I suppose I would try it on a higher end game or the demo's in GPU Caps Viewer and see what happens.  If it causes problems then you have two reasons.  Otherwise I would just say it fails under load.
Sam

[dolfcao]

i have such problem too, gpu dead while using cgminer, don't know how to solve.

[seocrow]

I had a sick GPU

I found out it was overheating so resolved by adding 2 additional GPU fans in the back of my case to reduce the GPU temperatures.

I also reduced my intensity to 13 and got much better results.

[Wilderness]

as an update to my SICK gpu, I put it in my main system and tried some mining on it - SICK again. Then I thought I'd try some gaming and mid game I am getting flickering white lines occasionally so I'm thinking that in this instance it probably is faulty. Will try RMA. Not a big fan of power colour, wish I'd just got more of the gigabyte cards which seem to run for days and much cooler

[MashRinx]

I've never had to RMA a GPU yet so I don't know.  I suppose I would try it on a higher end game or the demo's in GPU Caps Viewer and see what happens.  If it causes problems then you have two reasons.  Otherwise I would just say it fails under load.
Sam

I have RMA'd a couple GPUs and agree with os2sam's suggestion.  Confirming that the card has issues in multiple use-cases will help.  I am sending a 7970 back to Gigabyte tomorrow actually, and although my issue is a bit more pronounced (no video at all as the primary and card isn't picked up by windows as secondary or tertiary (I love that word) card), my conversation with their support dept (required before getting a RMA) was < 5 mins.  I think the it fails after a short period of time under load" is a fair stance.

[2x0ninja]

Yeah that sounds like something you can RMA it for. Although the issue could be as simple as bad thermal paste/pad application at the factory. If your warranty allows you to remove the stock cooling it couldn't hurt to reapply thermal paste and give it a go, might save you some down time. Most decent aftermarket paste is way better then what they ship it with so you'd probably end up with better temps as well. If the card is already out of warranty you can pop it in the oven at 385 for 8-12 minutes to remelt and fix any failing solder.

[Wilderness]

I got a new card replaced under warranty and it works fine now, both as a single card and in a 4 card rig.

good tip replacing the thermal paste, ive done that for my second hard cards I've got but glad I got this one replaced under warranty