[SOLVED] PC restarting - advice on next steps

May 1, 2020
4
0
10
Hi guys,
I have returned from a year abroad to my PC which now seems to be cutting power when gaming, GPU load is almost never more than 60% as per the AMD system monitor, and CPU and GPU temps are 30-50 Celsius. Updating drivers does not help. Computer also restarts on GPU stress test, not on CPU stress test (both done with Furmark).
I understand from looking at the forums that the only way to understand whether this is really a PSU issue vs GPU or Mobo is to replace the PSU. I wanted to make sure this is correct and get some advice on what the replacement PSU should be, or whether another approach is more appropriate.
Any help is much appreciated!
Cheers

Specs:
CPU: Intel Core i5 4460
GPU: Sapphire Radeon R9 390
Mobo: ASRock H97M Anniversary
PSU: XFX XTR 550
 
Last edited:
Solution
I'd read GPUs do not have constant power consumption, so pulling more power during a stress test from a dodgy PSU for an extended period of time can make it trip. Is this wrong?
Maybe wishful thinking since obviously it's the more expensive component to replace... shame to hear about its reputation.

See, this can happen but it's not 'extender period of time' related. Some AMD GPUs are notorious for pulling extreme amounts of power in very short time periods (as in the matter of milliseconds, it's called transient load), like twice as much as average power draw. This is enough to trip PSU's OCP tho if it's sensitive enough. And despite this PSUs isn't dodgy, it's made by Seasonic, an analog to their G series, i think i heard...

Eugene$$

Reputable
Dec 15, 2019
179
30
5,290
It could be a gpu issue or ram, its usually one of them, can you borrow another gpu from a friend? have you tried to reset the bios? have you tried to replug the gpu? also try a memfest test..
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
GPU load is almost never more than 60% as per the AMD system monitor, and CPU and GPU temps are 30-50 Celsius.

And how is this related to PSU 'cutting power' ? There's no such thing, it either provides enough power and PC works or ot doesn't and PC shuts down. Shutdown under stress test may be the case of tripping OCP per se but it's somewhat unlikely, it's probably just your GPU being unstable, it's an old GPU and this lineup are known to have 'cold solder' problems.
 
May 1, 2020
4
0
10
It could be a gpu issue or ram, its usually one of them, can you borrow another gpu from a friend? have you tried to reset the bios? have you tried to replug the gpu? also try a memfest test..

Thanks for this - tried a memtest and everything came back normal, difficult to borrow a gpu in quarantine but will definitely try resetting bios and replugging GPU.
 
May 1, 2020
4
0
10
And how is this related to PSU 'cutting power' ? There's no such thing, it either provides enough power and PC works or ot doesn't and PC shuts down. Shutdown under stress test may be the case of tripping OCP per se but it's somewhat unlikely, it's probably just your GPU being unstable, it's an old GPU and this lineup are known to have 'cold solder' problems.

I'd read GPUs do not have constant power consumption, so pulling more power during a stress test from a dodgy PSU for an extended period of time can make it trip. Is this wrong?
Maybe wishful thinking since obviously it's the more expensive component to replace... shame to hear about its reputation.
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
I'd read GPUs do not have constant power consumption, so pulling more power during a stress test from a dodgy PSU for an extended period of time can make it trip. Is this wrong?
Maybe wishful thinking since obviously it's the more expensive component to replace... shame to hear about its reputation.

See, this can happen but it's not 'extender period of time' related. Some AMD GPUs are notorious for pulling extreme amounts of power in very short time periods (as in the matter of milliseconds, it's called transient load), like twice as much as average power draw. This is enough to trip PSU's OCP tho if it's sensitive enough. And despite this PSUs isn't dodgy, it's made by Seasonic, an analog to their G series, i think i heard that this platform are prone to tripping with AMD Vega which is notorious for high transient load spikes, so if R9 390 has high transient load spikes too (which i'm not sure about) that would explain this. You need to get a different PSU temporarily to test this theory, smth decent at least.
 
Solution
May 1, 2020
4
0
10
See, this can happen but it's not 'extender period of time' related. Some AMD GPUs are notorious for pulling extreme amounts of power in very short time periods (as in the matter of milliseconds, it's called transient load), like twice as much as average power draw. This is enough to trip PSU's OCP tho if it's sensitive enough. And despite this PSUs isn't dodgy, it's made by Seasonic, an analog to their G series, i think i heard that this platform are prone to tripping with AMD Vega which is notorious for high transient load spikes, so if R9 390 has high transient load spikes too (which i'm not sure about) that would explain this. You need to get a different PSU temporarily to test this theory, smth decent at least.

This is helpful, thanks. Just today I've noticed that the GPU also crashes immediately after exiting, even if I run it for a second and it's still cool, which I suppose also suggests it's unstable. Will try and source an alternative to test.