Need Emergency Advice Please!!

[mx19]

So I went to check my miner online and noticed it had been offline for 45 minutes, i went to start it again and could smell burning, turned it on and the first gpu caught fire with like a torch like flame where the power connectors are attached!!! Has anyone heard of this? Why might it have happened?!

We are running 4 gigabyte7970's at around 75c and 673kh/s each In an open custom built frame. All the GPU's are on non-powered risers, could this have been an issue? we have a 1500watt psu running them.

Please could i have some advice, this is pretty scary as i also live on the 66th floor of a high rise building and it could have taken the whole thing down!!

[Starscream]

So I went to check my miner online and noticed it had been offline for 45 minutes, i went to start it again and could smell burning, turned it on and the first gpu caught fire with like a torch like flame where the power connectors are attached!!! Has anyone heard of this? Why might it have happened?!

We are running 4 gigabyte7970's at around 75c and 673kh/s each In an open custom built frame. All the GPU's are on non-powered risers, could this have been an issue? we have a 1500watt psu running them.

Please could i have some advice, this is pretty scary as i also live on the 66th floor of a high rise building and it could have taken the whole thing down!!

Well, first off, let me reassure you that your rig on fire wouldn't take the whole building down ( I think  )

Second, as for the cause of the fire, there could be numerous reason for it, from which the most likely one is that your VRAMs overheated, as those can run very hot. Could also be a simple manufacturing defect, in which case you just got unlucky. I hope you still got warranty.

As far as the risers go, I doubt those are at fault since if it was an issue with non powered risers than your motherboard would have fried, or the pcie connectors, but not the GPU.

[anderl]

This can happen if you use molex to get 3 or 4 port power cords. Otherwise it could just be a manufacturing defect.

[Caiapfas]

let's see pictures. sounds amazing!

[tokyoghetto]

my msi 7950 windforce caught fire once. Its due to a shitty VRM design. RMA'd that card right away. 3 year warranty. Life is good.

[kalus]

pics?

[Wipeout2097]

Did you undervolt the cards?

[LostDutchman]

So I went to check my miner online and noticed it had been offline for 45 minutes, i went to start it again and could smell burning, turned it on and the first gpu caught fire with like a torch like flame where the power connectors are attached!!! Has anyone heard of this? Why might it have happened?!

We are running 4 gigabyte7970's at around 75c and 673kh/s each In an open custom built frame. All the GPU's are on non-powered risers, could this have been an issue? we have a 1500watt psu running them.

Please could i have some advice, this is pretty scary as i also live on the 66th floor of a high rise building and it could have taken the whole thing down!!

Lack of powered risers most likley.

My $.02.

[AtlantisPlatform]

So I went to check my miner online and noticed it had been offline for 45 minutes, i went to start it again and could smell burning, turned it on and the first gpu caught fire with like a torch like flame where the power connectors are attached!!! Has anyone heard of this? Why might it have happened?!

We are running 4 gigabyte7970's at around 75c and 673kh/s each In an open custom built frame. All the GPU's are on non-powered risers, could this have been an issue? we have a 1500watt psu running them.

Please could i have some advice, this is pretty scary as i also live on the 66th floor of a high rise building and it could have taken the whole thing down!!

Lack of powered risers most likley.

My $.02.

Guaranteed it had NOTHING to do with non powered or powered risers.  People just do not understand these damn risers and what function the powered part plays in the whole scheme of things. Of all the things that may have caused the fire, risers played 0 part in it.

[mx19]

Did you undervolt the cards?

No i haven't undervolted the cards at all, Will this have played a part?

On further inspection it looks like it it the VRM closest to the power connectors that has burnt...

[bjalbert]

Maybe dust went up in flames? I once had a soda fall through the top of my antec 1200 case and man I never smelt destruction like that before. Video card still stinks of blown capacitors and whatnot.

[mx19]

It is actually exactly the same as in this thread https://bitcointalk.org/index.php?topic=235112.new#new But they never said why it was caused... i have 4 of these and would love to stop it happening on another...

[Wipeout2097]

Did you undervolt the cards?

No i haven't undervolted the cards at all, Will this have played a part?

On further inspection it looks like it it the VRM closest to the power connectors that has burnt...

Indeed it did play a part. Undervolt them now. http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/

Look at the temperatures with GPU-z

[bjalbert]

I got 3 cards running hot and haven't seen it. But 80C=176F and that is at the sensor. I can see fine dust going boom kinda like a mining explosion no pun intended.

[mx19]

I have been trying to flash the bios with lower voltages, but they dont seem to work, do you have any idea what voltage selection i should be using?

[Wipeout2097]

I have been trying to flash the bios with lower voltages, but they dont seem to work, do you have any idea what voltage selection i should be using?

1080 mV as starting point, but some chips can go as low as 950 mV. For example

1075 mV -> 1100 Mhz
1050 mV -> 1050 Mhz
1020 mV -> 1000 Mhz
980 mV -> 950 Mhz

Currently I have:

7870XT: Tahiti LE : 0.95V -> 1000 Mhz
R9-270:Pitcairn: 1.05V -> 1100 Mhz
7970: Tahiti: 1.05V -> 1030 Mhz (too high voltage, but needed for MemoryCoin and VertCoin)
6950: Cayman: 1.1V -> 900 Mhz

Of course, the lower voltage you can reach for the particular core clock that maximizes hash rate, the better