GPU is not getting enough power

[ThriftStoreEnjoyer]

I have a really old Rx 570 XFX GPU which I assume is finally nearing its end.

The problem started more than a year ago when the system would crash whenever the GPU would draw too much power, at around 80 Watts. I took the GPU apart and cleaned the dust gathered around the VRMs. That fixed the problem and it did not repeat until recently. 
Now the same problem occurs: The system restarts when the GPU is under load.
I have cleaned the GPU thoroughly and inspected it for damage. Nothing caught my eye.

I tested the GPU in another system and got the same problem. This rules out the PSU or any other hardware.

Temps are stable, only peaking at 70 Celcius. I can only assume the GPU is somehow failing at drawing the required power.

 

It would be helpful if anyone could provide some insight into what I could do next to eliminate this problem.
Since I can't locate any damage or dust, I'd assume there are still some steps I can take.

But again, I don't have sound knowledge when it comes to Computers.

 

System Specs:

CPU: AMD Ryzen 5 2600
GPU: Rx 570 XFX
Motherboard: Micro-Star International Co. Ltd. B450M PRO-M2 MAX
PSU: Antec 500 Watts
RAM: 2x8 GB Dual-Channel DDR4 @ 1332MHz (Kingston)

SSD: 240 GB ADATA SU630

[Jurrunio]

Maybe the VRM is just barely adequately cooled? And as the components age, they are less efficient and therefore hit shutdown temperature easier.

 

Try use Hwinfo and see if it can pick up extra temperature sensors onboard. Typical OC software reads only the GPU core temperature

[KidKid]

13 minutes ago, ThriftStoreEnjoyer said:

I have a really old Rx 570 XFX GPU which I assume is finally nearing its end.

The problem started more than a year ago when the system would crash whenever the GPU would draw too much power, at around 80 Watts. I took the GPU apart and cleaned the dust gathered around the VRMs. That fixed the problem and it did not repeat until recently. 
Now the same problem occurs: The system restarts when the GPU is under load.
I have cleaned the GPU thoroughly and inspected it for damage. Nothing caught my eye.

I tested the GPU in another system and got the same problem. This rules out the PSU or any other hardware.

Temps are stable, only peaking at 70 Celcius. I can only assume the GPU is somehow failing at drawing the required power.

 

It would be helpful if anyone could provide some insight into what I could do next to eliminate this problem.
Since I can't locate any damage or dust, I'd assume there are still some steps I can take.

But again, I don't have sound knowledge when it comes to Computers.

 

System Specs:

CPU: AMD Ryzen 5 2600
GPU: Rx 570 XFX
Motherboard: Micro-Star International Co. Ltd. B450M PRO-M2 MAX
PSU: Antec 500 Watts
RAM: 2x8 GB Dual-Channel DDR4 @ 1332MHz (Kingston)

SSD: 240 GB ADATA SU630

Probably just an old GPU thing, but I'm not sure.

[emosun]

22 minutes ago, ThriftStoreEnjoyer said:

I tested the GPU in another system and got the same problem. This rules out the PSU or any other hardware.

Well no it just means the conditions are correct on both machines to manifest as the same problem. 

 

Could be psu related but the gpu is so old now it begs the question of how much of your time you're going to waste on it vs just finally replacing it

[ThriftStoreEnjoyer]

47 minutes ago, Jurrunio said:

Maybe the VRM is just barely adequately cooled? And as the components age, they are less efficient and therefore hit shutdown temperature easier.

 

Try use Hwinfo and see if it can pick up extra temperature sensors onboard. Typical OC software reads only the GPU core temperature

Temperatures are stable throughout the GPU, barely hitting 60 degrees Celcius before the system restarts.
GPU voltage was at 1V and VRM at 11.4V at the time of the crash. Drawing 90 Watts.

I should add that the crash is almost instant. Slowly raising the power and clock on MSI afterburner and at one point the screen goes black abruptly, leaving no time for temperatures to spike.

 

It is sort of like when you overclock a GPU well beyond its limits and it says no.

 

 

[ThriftStoreEnjoyer]

38 minutes ago, emosun said:

Well no it just means the conditions are correct on both machines to manifest as the same problem. 

Not sure I follow.

Only this particular GPU was raising problems in both systems.

[emosun]

3 hours ago, ThriftStoreEnjoyer said:

Not sure I follow.

Only this particular GPU was raising problems in both systems.

Well the two computers you tested the gpu in are not representative of every computer on earth. It's just an exercise in not concluding in absolutes. The gpu is not functional in the two computers you've tried aka the two computers that matter to you. So while the gpu could still be functional,  to you specifically it's functionally dead.

 

But again it doesn't matter why it's broken , just that it is. So time to change it anyway

[Real_Smoky]

On 1/29/2025 at 11:49 PM, ThriftStoreEnjoyer said:

I have a really old Rx 570 XFX GPU which I assume is finally nearing its end.

The problem started more than a year ago when the system would crash whenever the GPU would draw too much power, at around 80 Watts. I took the GPU apart and cleaned the dust gathered around the VRMs. That fixed the problem and it did not repeat until recently. 
Now the same problem occurs: The system restarts when the GPU is under load.
I have cleaned the GPU thoroughly and inspected it for damage. Nothing caught my eye.

I tested the GPU in another system and got the same problem. This rules out the PSU or any other hardware.

Temps are stable, only peaking at 70 Celcius. I can only assume the GPU is somehow failing at drawing the required power.

 

It would be helpful if anyone could provide some insight into what I could do next to eliminate this problem.
Since I can't locate any damage or dust, I'd assume there are still some steps I can take.

But again, I don't have sound knowledge when it comes to Computers.

 

System Specs:

CPU: AMD Ryzen 5 2600
GPU: Rx 570 XFX
Motherboard: Micro-Star International Co. Ltd. B450M PRO-M2 MAX
PSU: Antec 500 Watts
RAM: 2x8 GB Dual-Channel DDR4 @ 1332MHz (Kingston)

SSD: 240 GB ADATA SU630

 

On 1/30/2025 at 4:44 AM, emosun said:

Well the two computers you tested the gpu in are not representative of every computer on earth. It's just an exercise in not concluding in absolutes. The gpu is not functional in the two computers you've tried aka the two computers that matter to you. So while the gpu could still be functional,  to you specifically it's functionally dead.

 

But again it doesn't matter why it's broken , just that it is. So time to change it anyway

Assuming the other two computers are known to be good (i. e. stable with another GPU), testing the 570 in both of them and having the system crash in a similar way is as conclusive as it gets that the 570 is the problem.

The graphics card is old and some SMD might have a poor contact or the GPU itself is degrading. The only thing you can do now double-check if the thermal pads are contacting the VRAM chips because the GPU might not report VRAM temps. Otherwise, you could always try baking it or treating it with a heat gun for a couple more weeks of usage, but probably not much longer than that. 

The graphics card is old, some of them may survive for a couple more years, but they are starting to degrade. 

I have some experience with XFX cards, the ones I got a few years ago were likely used for mining but died very early on. Might be that that particular series of RX 470, 480, 570 and 580 cards was not that great from XFX.