Question Extreme perfomance issuses. What is broken?

Feb 6, 2021
5
1
15
0
I need help to figure out what is broken.

CPU: AMD FX 8350
GPU: NVIDIA GeForce GTX 970
Motherboard: ASRock 970 Extreme3 R2.0
RAM: 2 * 4GB DDR3

Around 2 months ago my PC would not turn on, I checked the power plug and tried again.
In that moment my PC turned on but there was a quite sound (maybe like a capacitor burn out?).
I started a game to see if everything is working as it should, it is not.

Whatever game I play, after 30 seconds the FPS drop down to 10 and go back up after 10 seconds,
10 seconds later the frames drop again and the cycle repeats.
Every game is unplayable (...except among us) even with low settings.
I tried paradox games (Victora 2, HOI4, EU4) and they are all the same.

Normal use mostly works like Chrome, Zoom and other desktop applications.
But I encountered the same problem when I was using Discord (with webcam) and watchtogther in Chrome,
meaning extreme performance problems. In that moment I sound like a roboter in Discord.

I just cant figure out what the problem is, Is it the CPU? the GPU? or maybe the motherboard?

I did a stress test with my CPU (prime95), the cpu clock drops every few seconds. The temperatures are fine.
I did a benchmark with my GPU (unigine valley) and whenever it stutters, the clock speed of the CPU also drops. Idk if that is normal.

Here is the clock speed in HWMonitor during the CPU stress test:


Here is the behaviour in taskmanager during the stress test:


What could be the problem and what can I do to find out what component the problem maker is?
 

Phaaze88

Titan
Ambassador
1)The cpu thermals are an unknown, because you're using the wrong app to read them.
AMD Overdrive here: https://www.techspot.com/downloads/4645-amd-overdrive.html
Check core thermals again.
Warning It will confuse you - like many others who weren't familiar with the app, so to make it easier for you:
Double digits: good
Single digits: getting a bit toasty, but not really favorable
0: Bad, thermal throttle
Negative values: Also bad

2)Weak 4+1 phase VRM mobo paired with an FX-8350. That was bound to go wrong sooner or later.
There's many of those kinds of 125w FX cpu + weak mobo VRM threads here: they burn out faster because they can't actually handle those 125w cpus for very long.
You'll likely need a new mobo, but no one here is going to tell you to get another AM3+ board without a darn good reason; that would be a waste of money.
Time for a new cpu/mobo/ram combo.
 
Feb 6, 2021
5
1
15
0
1)The cpu thermals are an unknown, because you're using the wrong app to read them.
AMD Overdrive here: https://www.techspot.com/downloads/4645-amd-overdrive.html
Check core thermals again.
Warning It will confuse you - like many others who weren't familiar with the app, so to make it easier for you:
Double digits: good
Single digits: getting a bit toasty, but not really favorable
0: Bad, thermal throttle
Negative values: Also bad

2)Weak 4+1 phase VRM mobo paired with an FX-8350. That was bound to go wrong sooner or later.
There's many of those kinds of 125w FX cpu + weak mobo VRM threads here: they burn out faster because they can't actually handle those 125w cpus for very long.
You'll likely need a new mobo, but no one here is going to tell you to get another AM3+ board without a darn good reason; that would be a waste of money.
Time for a new cpu/mobo/ram combo.
Thanks for the answer.
1) The temperature reserve is at around 40, when I do a stress test it goes down to ~30.
Around there the clock speed drops to 1380MHz and the temperatur reserve goes back to 40 - 50.
I assume the temperature is okay then?

2) Yes I know, found that out a couple months ago while having perfomance issues.
Obviously there is no point in buying a new AM3+ mainboard and I am willing to buy a new cpu/mobo/ram combo.
But I want to make sure that will fix it and that its not actually the gpu that is broken.
 

Phaaze88

Titan
Ambassador
1)Yes and no: It throttles before you can even put a decent load on it... that's no good.

2)The damage may have already been done, but the only other thing I could think you could try:
-remove the front, top, and side panels of the chassis
-get a table or box fan
-aim that fan inside the chassis
-check thermals/frequencies again
 
Feb 6, 2021
5
1
15
0
It took off the heat sink to have a look at the MOSFETs of the VRM, could that one that is slighty discoloured be the reason? 3rd from the top right side

 

ASK THE COMMUNITY