Used gpu off a miner DOA?

iBSOD · 15 Jun 2018 at 14:51

Hi there,

Received a RX 580 Strix I bought off a miner today, but appears it could be DOA.

Under load the card crashes immediately, system freezes (hard hang) fans spin at 100 and system requires a hard reboot to get going again. I can get the system to last more than 2 seconds by lowering the power limit of the card in wattman, but still crashes within a minute under high load. If I lower gfx settings also the system lasts longer before crashing. PSU is a RM850i.

What do you guys think it could be? The card was just sent wrapped in bubble wrap with no anti-static bag, maybe its ESD damage to the card?

Thanks

Kelt · 15 Jun 2018 at 14:55

Borked card.

Send it back for a refund.

Armageus · 15 Jun 2018 at 14:57

Unlikely to be ESD imo - I've received lots of graphics cards and processors over the years that haven't been in anti-static bags with no issues.

Sounds more like either DOA, or flashed to a very specific mining BIOS that your board doesn't like for whatever reason.

iBSOD · 15 Jun 2018 at 15:03

Armageus said:
Unlikely to be ESD imo - I've received lots of graphics cards and processors over the years that haven't been in anti-static bags with no issues.

Sounds more like either DOA, or flashed to a very specific mining BIOS that your board doesn't like for whatever reason.

The seller flashed the stock BIOS back on the card before sending it to me.

The seller agreed to send me a replacement if this ones dead

Thekwango · 15 Jun 2018 at 15:05

iBSOD said:
The seller flashed the stock BIOS back on the card before sending it to me.

The seller agreed to send me a replacement if this ones dead

so there's no real question tbh, the card is borked, get them to replace.

Pocah · 15 Jun 2018 at 15:59

Sounds like heat to me. Cards can go a number of ways and one of those ways is that it collapses as soon as some component reaches a specific temperature. Eventually it gets worse, the failure temperature gets lower and lower then one day the card dies completely. I am afriad you need to send it back :/

iBSOD · 15 Jun 2018 at 16:02

Pocah said:
Sounds like heat to me. Cards can go a number of ways and one of those ways is that it collapses as soon as some component reaches a specific temperature. Last one I saw would collapse at exactly 70 degrees on the dot. Eventually it gets worse, the failure temperature gets lower and lower then one day the card dies completely. I am afriad you need to send it back :/

Card stays under 55c at load, but for some reason the fans spin up to >90% at load (before it dies) despite temps being very low. I'm guessing this fan profile is made fro mining, but I don't get how that profile is still there after flashing stock bios and in a different pc etc

JediFragger · 15 Jun 2018 at 16:02

Don't forget to get him to cover the postage back.

Pocah · 15 Jun 2018 at 16:42

Armageus said:
Unlikely to be ESD imo - I've received lots of graphics cards and processors over the years that haven't been in anti-static bags with no issues.

Sounds more like either DOA, or flashed to a very specific mining BIOS that your board doesn't like for whatever reason.

That's a bit like saying "I've crossed the road loads of times over the years and never been run over". Static can not only kill a product it can leave it apparently just fine but reduce it's life. So you never really know if you have had a problem or not.

iBSOD said:
Card stays under 55c at load, but for some reason the fans spin up to >90% at load (before it dies) despite temps being very low. I'm guessing this fan profile is made fro mining, but I don't get how that profile is still there after flashing stock bios and in a different pc etc

It's doesn't have to be the GPU that is at fault, it could be a host of different components on the board to blame.

As to why the fan profile is stuck I am not well-enough versed in AMD cards to say why, but I would have to say that it's too much of a coincidence that the fan is having issues and so apparently is the board. I suspect the problems are one and the same... not that it makes much difference because I still think the card is shot. And also you want to be very careful here, if you tell the seller you flashed the BIOS he may say it's your fault it's not working.

MrRockliffe · 15 Jun 2018 at 16:52

Pocah said:
That's a bit like saying "I've crossed the road loads of times over the years and never been run over". Static can not only kill a product it can leave it apparently just fine but reduce it's life. So you never really know if you have had a problem or not.

True, but many new components have ESD protection, and can handle the higher voltages. I've seen a few videos of people trying to break stuff with ESD and have failed.

I don't think it's much of a problem these days.

Vince · 15 Jun 2018 at 17:19

Anybody remember ecs elitegroup boards? You only had to look at them in the wrong way and they would die

update the bios on your board, test again then send back :cool:

if the rx580 has dual bios flip the switch and have a go on the spare bios!

AiiR · 15 Jun 2018 at 17:59

It's a mining card, what more did you expect.

Hope you got a bargain to buy something that's been running 24/7 in unknown conditions.

0h_no_not_again · 15 Jun 2018 at 18:53

MrRockliffe said:
True, but many new components have ESD protection, and can handle the higher voltages. I've seen a few videos of people trying to break stuff with ESD and have failed.

I don't think it's much of a problem these days.

Funnily enough newer CMOS based devices are extremely sensitive to ESD. Way way more than previous generations of TTL and >100nm CMOS geometries.
On the topic of ESD protection, very few devices include it, due to clamping diodes taking up space and impacting signal integrity. Those that do are often rated to 2kV, which is actually very low in the scheme of ESD. A good audible zap to your finger is often 15kV.
Devices that are connected to external IO usually have some form of protection, but I guarantee you the DDR, GPU, VRM FET gates, VRM controllers, etc do not.

The worst issue is, an ESD event can cause a host of failures that is usually slight degradation but sometimes you'll get random behavior and even short circuits. So all those DOA/flakey new pieces of hardware....
I've seen a failure where one a FET had a 300Ohm resistance between the gate and the source. This would cause a catastrophic failure of a VRM, or some odd things to happen if it were on a DDR data/address line.

Anyway, ESD is easy enough to avoid - constantly ground yourself to your case while it is plugged in (switched off) / wear a wrist strap. Avoid work in a low humidity environment usually caused by AC or heating in winter.

Vince · 15 Jun 2018 at 19:57

AiiR said:
It's a mining card, what more did you expect.

Hope you got a bargain to buy something that's been running 24/7 in unknown conditions.

People say this all the time but seriously my old 5850's ran for literally 3 years flat out at like 90 degrees plus. I've still got them and apart from grinding fans and non existent drivers they work perfectly. I've seen machines that literally have so much dust in them that the fans aren't moving and the things just carry on working, I bet most people have an old p4 or overclocked barton laying around that you wouldn't think of using these days but you can pretty much guarantee that if you plug it in it will probably work.

Pocah · 15 Jun 2018 at 22:15

MrRockliffe said:
True, but many new components have ESD protection, and can handle the higher voltages. I've seen a few videos of people trying to break stuff with ESD and have failed.

I don't think it's much of a problem these days.

Pah they are amateurs! I can destroy any card in five seconds flat, and in fact I did so, much to my bosses dismay. But yes ESD protection usually protects a card up to about the point where you can feel the jolt. If you can feel it, then you have damaged it. Whether it's fatal or you just reduced it's life is another matter. But in general given the high cost of hardware it's not a good idea to ignore ESD.

varkanoid · 15 Jun 2018 at 22:27

Makes me laugh the talk of ESD when its been used for mining and thats the unknown and more likely factor. Strictly speaking its not DOA either as it does work in a small capacity. Just put it down to the OP lost the lottery and return it as faulty.

(I bought an AMD 290 ex mining card, cooler and card caked in some kind of black sticky gunk. Cleaned it all off replaced the cooler and worked fine, £80 that was with another working card albeit no cooler so not all bad. Worth a punt sometimes)

Sasahara · 15 Jun 2018 at 22:49

One problem with ex-mining card can be they haven't ever been used for gaming.

So a fault that was present when it was new may not show itself when used for mining, but as soon as it's used as a gaming card the fault becomes apparent.

iBSOD · 19 Jun 2018 at 10:35

So after trying the GPU in another computer its perfectly stable! Wreckon my power supply is faulty? I'm gonna try another pcie cable when I get back home

Clarkson · 19 Jun 2018 at 11:06

^^ But..But...dem miners! :mad:

(

)

iBSOD · 19 Jun 2018 at 17:34

Any ideas what might be wrong with my PC? Been trying for ages to get it to run

When running Heaven the card stays clocked around 500-900mhz despite temps being cool and FPS being low 30s. This didn't happen when in another pc, it stayed around 1360mhz in the same test.