Ex-mining GPU question!

[ppanickk]

Hello everyone

I have purchased an ex-mining Sapphire Nitro RX480 8GB graphics card from an eBay seller on a for parts or not working basis.

The card was in a sorry state when I received it, it was filthy and quite corroded. Since then, I have cleaned it up and refurbished it with new thermal pads and thermal paste. I have also flashed the original gaming bios on it as the previous owner had a mining bios on it that prevented it from working correctly.

The issue is, despite my efforts, the card will freeze the system when running a benchmark or a game. When cold, it might run for a bit longer, but it will still (hard) freeze the system after a while.

So the most likely scenario is that the card has deteriorated from prolonged mining use to a point it cannot support the stock clocks any more. I'm thinking to lower the core / memory clocks in an attempt to rectify the issue but perhaps a (modified) bios would be a better solution? What do you think? The card has dual bios so there is no risk of bricking it.

Any help or advice is highly appreciated.

Best
Ppanickk

 

[Macro Device]

Well, if we don't know how it behaves downclocked we also don't know if it makes sense to downclock it and render it permanent via BIOS.

Are all temperatures in check? Got a technician to test resistances, voltages, etc?

Since it's a mining-worn GPU my main suspicion falls upon two things, VRAM and VRM. Test ~1.7 (6.8 effective) GHz VRAM first, lower power limit (-20%) second.

If that's VRAM then just run it downclocked and call it a day. If it's VRM it could make sense to replace it if you got compatible spares in OK condition in a close enough proximity.

 

[LabRat 891]

When cold, it might run for a bit longer, but it will still (hard) freeze the system after a while.

If it were my card, it'd be getting a bake.
That sounds like cold solder joints.


Or,
'worn out' power components, no longer operating within spec. Which, isn't worth 'fixing'.

 

[VuurVOS]

The issue is, despite my efforts, the card will freeze the system when running a benchmark or a game. When cold, it might run for a bit longer, but it will still (hard) freeze the system after a while.

The VRM is busted and needs to be replaced. A temporary fix would be to add thermal pads on the backside of the VRM so heat can be absorbed by the backplate.

https://www.reddit.com/r/Amd/comments/5sqog2

 

[AleXXX666]

Hello everyone

I have purchased an ex-mining Sapphire Nitro RX480 8GB graphics card from an eBay seller on a for parts or not working basis.

The card was in a sorry state when I received it, it was filthy and quite corroded. Since then, I have cleaned it up and refurbished it with new thermal pads and thermal paste. I have also flashed the original gaming bios on it as the previous owner had a mining bios on it that prevented it from working correctly.

The issue is, despite my efforts, the card will freeze the system when running a benchmark or a game. When cold, it might run for a bit longer, but it will still (hard) freeze the system after a while.

So the most likely scenario is that the card has deteriorated from prolonged mining use to a point it cannot support the stock clocks any more. I'm thinking to lower the core / memory clocks in an attempt to rectify the issue but perhaps a (modified) bios would be a better solution? What do you think? The card has dual bios so there is no risk of bricking it.

Any help or advice is highly appreciated.

Best
Ppanickk

advice 1: don't buy very old gpus which are ex-mining era. Potential scam, and potential brick.
advice 2: ex-mining purchases this time MUST start with GTX 16, RTX 20 series, RX 5*** series. Though, Nvidias might be in better condition as most a way cooler lol.

 

[Macro Device]

RTX 20 series

Non-Super variants excluded. They are much more worn-out.

RX 5*** series

I'd rather not. Their behaviour under light loads is something... else.

Polaris GPUs, though, are very tough nuts to crack. My RX 480 still works and handles some OC despite it being ancient and used in mining. Their most major disadvantage is that they are weak and obsolete by today's standards.

 

[AleXXX666]

Non-Super variants excluded. They are much more worn-out.

I'd rather not. Their behaviour under light loads is something... else.

Polaris GPUs, though, are very tough nuts to crack. My RX 480 still works and handles some OC despite it being ancient and used in mining. Their most major disadvantage is that they are weak and obsolete by today's standards.

was happy to get RX 5700 and now sold it to friend who is also a happy user of it...

 

[ppanickk]

Well, if we don't know how it behaves downclocked we also don't know if it makes sense to downclock it and render it permanent via BIOS.

Are all temperatures in check? Got a technician to test resistances, voltages, etc?

Since it's a mining-worn GPU my main suspicion falls upon two things, VRAM and VRM. Test ~1.7 (6.8 effective) GHz VRAM first, lower power limit (-20%) second.

If that's VRAM then just run it downclocked and call it a day. If it's VRM it could make sense to replace it if you got compatible spares in OK condition in a close enough proximity.

Thanks for responding.

I do have one question about the power limit and the rationale of lowering it.

Initially, I was thinking to try the opposite really. If the stock core / memory clocks are unsustainable, perhaps with more power they can be sustained, no?

If it were my card, it'd be getting a bake.
That sounds like cold solder joints.


Or,
'worn out' power components, no longer operating within spec. Which, isn't worth 'fixing'.

A bake is an option but there is also the risk of completely bricking the card. Happened to me a few times.

The VRM is busted and needs to be replaced. A temporary fix would be to add thermal pads on the backside of the VRM so heat can be absorbed by the backplate.

https://www.reddit.com/r/Amd/comments/5sqog2

That would suck really as I am not good with the soldering iron. It is on my to do list though, one day!

advice 1: don't buy very old gpus which are ex-mining era. Potential scam, and potential brick.
advice 2: ex-mining purchases this time MUST start with GTX 16, RTX 20 series, RX 5*** series. Though, Nvidias might be in better condition as most a way cooler lol.

I got this for cheap just for the sake of playing with it. If it does end up working (doubtful), I will use it in one of my systems. If it doesn't, no harm is done.

 

[ir_cow]

I think it could be two things. 1: A Mosfets is going bad. Once up to temp it drops voltage causing the core to crash 2: The GPU been abused and requires more voltage for the same clocks. Could be both together.

On a side rant. It sucks you bought this card. Just gives miners more profit and encourages them to continue this practice.

 

[VuurVOS]

I think it could be two things. 1: A Mosfets is going bad. Once up to temp it drops voltage causing the core to crash 2: The GPU been abused and requires more voltage for the same clocks. Could be both together.

Its not abuse but poor mosfet quality. Friend of mine had a RX480 from Sapphire with the same issue. It occurred at the end of the warranty.
He was the first owner of the card and didn't sent the card for repairs because it only happend in one game....one year outside the warranty it got worse
The thermal pads mod for the VRM did the job for some time until a mosfet died