GPU issues (R9 390)

[jackoboy9]

Hi Guys,

 

So I was greedy and decided to go for an extra 50MHz on my core clock of my MSI R9 390. To get it to 1150MHz. It was perfectly stable at 1100MHz with +19mV on the core.

 

Anyways, 1150 was stable enough to run 3DMark 11 with no artefacts, but when I loaded up my miner, the monitor didn't detect a display input about 2 minutes into the mining process.

 

So I rebooted the PC and it would crash at the welcome screen. Me shitting myself, I attempted to boot into safe mode (which isn't easy on Windows 10 UEFI with an SSD) after multiple attempts, I got in. I used DDU to uninstall the drivers and then rebooted into normal windows.

 

Normal windows with no GPU drivers worked fine - no crashes. I try to install the latest Crimson drivers and it crashes during installation. I then reboot into safe mode, use DDU, then boot into normal windows and use the 'update driver' option in device manager, pointing it to the C:/AMD directory where the crimson drivers are extracted to.

 

Installs fine and everything is hunky dory.

 

I play a game for just over an hour and no crashes (at the overclocked 1130MHz core with +56mV). Later on, I boot up the miner and it crashes again. Did the whole process again and tried this morning with the 1100MHz core and it crashed again.

 

I have also tried the the stock 1040MHz core, and the same thing is happening!!!

 

Thanks for taking the time to read this.

[Dionyz]

Looks like you have burnt the card. Maybe try reinstalling the game, and see if that works.

[jackoboy9]

4 minutes ago, Dionyz said:

Looks like you have burnt the card. Maybe try reinstalling the game, and see if that works.

How is that possible? The card never went above 77C. I'm currently chilling in Windows without anything demanding running but the instant I open furmark or mine for about 15 seconds the screen loses the input. Windows is still running because sounds are still playing.

[App4that]

I would say it's unlikely you burnt the card. My suspicion it the mining software. If you have Rise of the Tomb Raider, play that. That game crashes easier than any I have with overclocks. If you can play for more than an hour then I'd try reinstalling the mining software.

[jackoboy9]

5 minutes ago, App4that said:

I would say it's unlikely you burnt the card. My suspicion it the mining software. If you have Rise of the Tomb Raider, play that. That game crashes easier than any I have with overclocks. If you can play for more than an hour then I'd try reinstalling the mining software.

Tried furmark, mining, GTA V. They all cause the GPU to cease giving an input to the monitor.

[App4that]

Damn. You could try reinstalling windows. Or contact MSI and RMA it. If it was me I'd try reinstalling windows first.

[jackoboy9]

5 minutes ago, App4that said:

Damn. You could try reinstalling windows. Or contact MSI and RMA it. If it was me I'd try reinstalling windows first.

I can't imagine it being a software problem. When the monitor loses an input, and I press the reset button, the motherboard beeps 1 long and 3 short. Then there is no display no matter how many times I press the reset button I'll just keep getting the beep code. The only thing that fixes it is to hold the power button (to turn the PC off) and then turn it back on again and for some reason that works fine and the GPU outputs a signal to the monitor.

[App4that]

Definitely a long shot, but RMA can take weeks. You could also try pulling RAM. And pulling the card and reinstalling it. Stranger things have worked for me.

[jackoboy9]

21 minutes ago, App4that said:

Definitely a long shot, but RMA can take weeks. You could also try pulling RAM. And pulling the card and reinstalling it. Stranger things have worked for me.

Just tried it. Reseating RAM, switching PCIe lanes. Nothing works.

 

In fact, when i switched PCIe lanes, the PC refused to display anything no matter what I tried. When I use the first slot, I can get a display when I cut the power and then turn it on again but that didn't work with the 2nd lane...

[App4that]

If at all possible see if you can try another card. But the fact you had issues in lane two is odd.

[jackoboy9]

12 hours ago, App4that said:

If at all possible see if you can try another card. But the fact you had issues in lane two is odd.

Ok so there's a full on burning smell coming from the GPU. Not the CPU/mobo/RAM/PSU. It's the GPU.

 

Just left furmark running for 5 minutes on 520MHz clock and 750MHz memory after a night's rest (lowest I can set it in afterburner) and it didn't seem to shut off but when I came back into the room, it stank of this burning smell. I feel like either one of the voltage regulators has gone, or something has found its way to the GPU and is dying and burning onto the PCB. This is only at 70C, so I dunno why it's happening, but jesus, did it smell!!!

[App4that]

RMA time, sorry. Any luck you'll get a 390x back.

[jackoboy9]

1 minute ago, App4that said:

RMA time, sorry. Any luck you'll get a 390x back.

It just came to me. It smells like an exploded diode, but yeah I feel an RMA is in order and since the retailer is out of stock of 390s I'll probably have to wait a month 

[jackoboy9]

So I thought the burning smell was from the GPU. Turns out the burning smell was from molten plastic around the 6 pin and 8 pin connectors. Now, I'm not sure whether the PSU/cables are faulty or the GPU is faulty or if they're both faulty, but the burning smell was not a capacitor, diode or VRM like I previously thought.

 

I only noticed this when I bought a spare GPU off eBay to tie me over in situations like these when I noticed the pin was melted on the 6pin connector. From the pictures, you can clearly see that the connector has melted on both cables. (Both 6+2pin adapters originate from the same cable).

 

I am angry to say the least because I spent a lot of money on the PSU so if it is indeed the PSU's fault then that's quite worrying.

Weirdly enough it happened to two pins on two separate splits. I already RMA'd my GPU which was received today, so I'll know within a week whether anything is wrong with it.

 

What do you guys think?

 

[SageOfSpice]

That screams PSU issue to me.

Under no circumstances should a PSU be outputting enough power to melt it's own connectors. That's a serious fault.

 

[YongKang]

Try another cable to see if the same thing occurs. 

[xAcid9]

On 6/17/2016 at 2:09 AM, jackoboy9 said:

Hi Guys,

-snip-

So I rebooted the PC and it would crash at the welcome screen. Me shitting myself, I attempted to boot into safe mode (which isn't easy on Windows 10 UEFI with an SSD) after multiple attempts, I got in. I used DDU to uninstall the drivers and then rebooted into normal windows.

-snip-

I guess your OC is not stable for mining. 
 

Also that boot loop you get was caused by Crimson/Radeon settings. 
When you OCed card crashed and reboot, Crimson will revert the voltage to default but keep the last OC value you set so if your 1150mhz is not stable at default voltage it will keep crashing on winlogon. It's not the card being faulty or anything.

If you want to find stable overclock for your card, install the last CCC driver which is 15.11 iirc?

 

That PSU cable totally farked up.

[jackoboy9]

15 minutes ago, YongKang said:

Try another cable to see if the same thing occurs. 

The other end is buggered too.

 

http://imgur.com/a/hzthw

[YongKang]

4 hours ago, jackoboy9 said:

The other end is buggered too.

 

http://imgur.com/a/hzthw

Well, brother, it's time for RMA or hit up customer support. 

[jackoboy9]

53 minutes ago, YongKang said:

Well, brother, it's time for RMA or hit up customer support. 

Yeah I've RMAd my PSU

[YongKang]

18 hours ago, jackoboy9 said:

Yeah I've RMAd my PSU

I would also check the mining program. The way  it put your psu into a bazam state is quite odd.

[jackoboy9]

On 07/07/2016 at 5:21 PM, YongKang said:

I would also check the mining program. The way  it put your psu into a bazam state is quite odd.

What could the mining program do to cause the PSU to overload?

[jackoboy9]

3 hours ago, Thunder87 said:

its not the PSU its just the PCI-E connector you're using just need to replace it. Here already looked for you. https://www.scan.co.uk/products/corsair-type-3-sleeved-black-pci-e-cable-cp-8920111

 

The connector female end (the PSU end) has been damaged and since my PSU is in warranty, I've RMAd it. Replacement should be arriving tomorrow.

[i_build_nanosuits]

On 2016-07-06 at 11:15 AM, jackoboy9 said:

What do you guys think?

 

 i would send those pictures to AMD

[YongKang]

29 minutes ago, i_build_nanosuits said:

 i would send those pictures to AMD

OMG... You troll, lol. I wonder what AMD gonna do, srsly?