RX 580 2048sp 8gb Problems

[VuurVOS]

After restarting pc flashing the RX570 vbios worked first try.

Then I restarted again bc it said so then it showed in gpu z as RX570 series.

Then I ran heaven benchmark, and it ran longer than usual to the point where I thought thats it then boom black screen.
I go sleep I will try the correct vbios tomorrow.

With the correct vbios? Your VRM is likely busted since it is unable to run at normal RX580 2048SP speeds.
Since the card is running slower than before, it takes more time before the VRM get (too) hot and then crashes.

Things you can do:

• Replace all thermal pads
• Put your fan at 100% all the time
• VRM thermal pads mod using the backplate as a heatsink (Different Sapphire RX 480 card but something like this: example)
• Find a E353 bios which runs at 1244MHz

 

[Jism]

Check / test the VRM temperatures first; bad pads, or even nothing, will cook them.

The card seems to hold up perfectly fine to the clocks, but likely shuts down due to VRM overheating (hardware protection).

 

[Ilija712]

Install Hwinfo. Run Furmark. Watch VRM temps of that RX580. I think they are cooking. Ideally you want them < 80 degrees. It's bad when they are above 80 degrees. There's ram chips next to it that are getting cooked because of it.

I had the same card; if cooled properly they can reach over 1650Mhz core.

I see I will focus on cleaning repasting repadding now.
Since you had same card do you remember the thermal pad thickness you used? Or can I just measure the thickness of the ones that are already there? I think the original ones are all a bit squished down thoe. And I dont think I took an image of it but theres a pad on the backplate aswell I will replace that too.

 

[Jism]

Yes it has both pads on top and the back of the card. The VRM's and esp through furmark can get hot.

When i ran with a hacked bios (the 250W version) i saw peaks of extreme furmark burning in going over 330W over a single 12V power connector.

They can be pushed; and the VRM can easily handle it. But they put quite the heat for such values.

On stock 60 degree should be the norm; not above 80. Also, keep in mind, the default fan profile is set to acceptable noise, so higher temps. The hot limit is 82 degrees.

I would always ramp up to 60% when you game or run a bench.

 

[AleXXX666]

Bios flashing is quite risky. I have done it before and I cannot boot with flashed bios. Need to revert back to normal. If you familiar with it is upon you whether thermal paste first or flashing first.

You can get any thermal paste, I prefer to use high-end thermal paste. For pads, you need to use the correct thickness. Ask the manufacturer what thickness is suitable for your needs. I ask powercolor and it said that i need 1.25mm, but its hard to get do I use 1.5mm instead

there is no "risk" if you flash OLD USED CHEAP stuff. It's not 7900 GRE, cmon...

 

[Ilija712]

Yes it has both pads on top and the back of the card. The VRM's and esp through furmark can get hot.

When i ran with a hacked bios (the 250W version) i saw peaks of extreme furmark burning in going over 330W over a single 12V power connector.

They can be pushed; and the VRM can easily handle it. But they put quite the heat for such values.

On stock 60 degree should be the norm; not above 80. Also, keep in mind, the default fan profile is set to acceptable noise, so higher temps. The hot limit is 82 degrees.

I would always ramp up to 60% when you game or run a bench.

Do you remember thermal pad thickness you used? I bought 1mm but the original ones seem to be 1.5mm but they are all squished and I think the squished part is closer to 1mm. Not sure thoe. Can you confirm the thickness?

 

[eidairaman1]

Do you remember thermal pad thickness you used? I bought 1mm but the original ones seem to be 1.5mm but they are all squished and I think the squished part is closer to 1mm. Not sure thoe. Can you confirm the thickness?

Try a feeler gauge

 

[Ilija712]

1mm feeler gauge fit 1.5mm didnt so I went and put 1mm but they are not making contact I can see a sliver of light between pad and plate. I will try and cut 1mm into 0.5 and add stack pads if temp same or worse I will buy 1.5mm

Its very hard to cut 1mm into 0.5 I ended up just stacking 1mm on 1mm. Tried running heaven benchmark and it was better I think. It ran a lot longer and the gpu clock was at like 600mhz that never happened before it would just run at highest constantly. It ran for like 5 mins after that fans reached 100% then I started benchmark and it went through half the scenes and turned off.

I think this is the first time ive seen this card run on lower gpu clock and the heaven benchmark was dropping frames and lagging. Maybe I messed something up during cleaning? Maybe the 2mm stacked pads?

I think I give up on this card, it was a fun experience trying to make it work but I will just sell it for parts.

Thanks to everyone for help and advice

 

[eidairaman1]

1mm feeler gauge fit 1.5mm didnt so I went and put 1mm but they are not making contact I can see a sliver of light between pad and plate. I will try and cut 1mm into 0.5 and add stack pads if temp same or worse I will buy 1.5mm

Its very hard to cut 1mm into 0.5 I ended up just stacking 1mm on 1mm. Tried running heaven benchmark and it was better I think. It ran a lot longer and the gpu clock was at like 600mhz that never happened before it would just run at highest constantly. It ran for like 5 mins after that fans reached 100% then I started benchmark and it went through half the scenes and turned off.

I think this is the first time ive seen this card run on lower gpu clock and the heaven benchmark was dropping frames and lagging. Maybe I messed something up during cleaning? Maybe the 2mm stacked pads?

I think I give up on this card, it was a fun experience trying to make it work but I will just sell it for parts.

Thanks to everyone for help and advice

Use the 1.5s, the proble is now, mining abuse kills them. Stick to brand new in box from a reputable source