Sign-in / Sign-up
Your question
Solved

How can I verify whether my PSU or my GPU is going bad?

Tags:
  • Power Supplies
  • Desktops
  • Homebuilt
  • Computers
  • Storage
  • GPUs
Last response: in Systems
June 24, 2014 9:49:30 AM

Earlier this year I built a computer pretty cheaply by scouting sales for like 2 months and using parts we had on hand. However now I'm having issues with it. It happens when my GPU goes under load and after a while the PC will just shut off. I suspect the power supply but it is strange that I can put my CPU under load and nothing but if I run furmark on the GPU for 3-5 minutes it goes kerflooey. It does not BSOD it is like a simple power failure. So my question is how can I eliminate one or both of these as a problem and/or verify if said hardware is bad?

PC Specs:
Generic 880 watt psu (already looking for a replacement but has done well in the past)
AMD A10 5800k CPU/APU (stock cooler) (evo 212 on the way)
ASRock fm2a75 pro4
2x8GB kingston hyperx 1600 DDR3
2x2GB G-skill 1600
GeForce GTX 275 (donation from brother)
OCZ Vertex3 120GB SSD
WD 1TB HDD

More about : verify psu gpu bad

June 24, 2014 10:05:17 AM

An 880 watt psu? That's a weird number, and I doubt highly that it's actually providing that much power. There are a few problems that give me red flags. The power supply definitely, but you're also using the stock cooler on that APU, and the stock coolers from AMD suck. You mention that you have a hyper 212 evo on the way...What I would do is wait for that to arrive, then run some tests again and see if you get shutdowns. I have a feeling you're pushing that little APU's graphics too hard on that stock cooler. But I also don't generally like the word "Generic" in a power supply, they tend to do things that aren't ideal, like claim 880 watts of power and make barely 300 watts, for instance. They also generally don't last very long, and also cannot keep themselves cool. The easiest thing to do at this point would be to get an app like core temp and check temperatures while benchmarking. That'll show you how hot things are getting while the benchmark is running. Power supplies can be tested by measuring voltages coming out of the lines, but it would be easier to check temps first.
m
0
l
June 24, 2014 11:19:47 AM

athlonman00 said:
An 880 watt psu? That's a weird number, and I doubt highly that it's actually providing that much power. There are a few problems that give me red flags. The power supply definitely, but you're also using the stock cooler on that APU, and the stock coolers from AMD suck. You mention that you have a hyper 212 evo on the way...What I would do is wait for that to arrive, then run some tests again and see if you get shutdowns. I have a feeling you're pushing that little APU's graphics too hard on that stock cooler. But I also don't generally like the word "Generic" in a power supply, they tend to do things that aren't ideal, like claim 880 watts of power and make barely 300 watts, for instance. They also generally don't last very long, and also cannot keep themselves cool. The easiest thing to do at this point would be to get an app like core temp and check temperatures while benchmarking. That'll show you how hot things are getting while the benchmark is running. Power supplies can be tested by measuring voltages coming out of the lines, but it would be easier to check temps first.


Core Temp does not work with APU's as the trinity APU utilizes a algorithm for how hot the CPU would be for the amount of work is being done. I have no clue how this actually works all I know is I hate it (not getting an actual accurate measure with a temperature probe or whatever). Core Temp gives me false readings such as 29 C. All things considered the temps should all be in check. The GTX 275 hits 80-82 degrees under full load which is acceptable but the computer shuts down before the gpu has a chance to throttle its self (it does so at 90 degrees)

Also since posting I have been getting BSOD stop 0xF4

Further more the computer is stock stock stock
stock clock on the CPU
stock clock on the GPU
stock clock on the Memory
m
0
l
Related resources
June 24, 2014 12:14:01 PM

So the temps seem to be ok? Assuming that temps are good and you're now getting a blue screen, means that now Windows itself can help you. Check event viewer for right around the last time it shut down on you. It'll have a lot of information about what's causing the blue screen and shutdowns. It's under control panel->Administrative tools in case you're not sure. Looking for anything with a red circle or yellow triangle, those are the bigger issues (Sorry if I sound belittling, I'm not sure your experience with PC's)
m
0
l
a b G Storage
a b ) Power supply
June 24, 2014 2:14:25 PM

Remove the gpu and use the apu. Put the pc under load and see if it has the same problem. If the same problem happens then it is the PSU for sure.
m
0
l
a c 1306 G Storage
a c 137 ) Power supply
a b B Homebuilt system
June 24, 2014 5:40:36 PM

The APU wont stress the PSU, its power is included in the 100w TDP of the processor. (ie - its not that powerful)

Test the card in another system if you can. If you need to guess, replace the PSU first in my opinion.
m
0
l
a b G Storage
a b ) Power supply
June 24, 2014 10:45:41 PM

popatim said:
The APU wont stress the PSU, its power is included in the 100w TDP of the processor. (ie - its not that powerful)

Test the card in another system if you can. If you need to guess, replace the PSU first in my opinion.


Wow, yeah, that's low power for an AMD chip. I agree on replacing the PSU first. Even if it isn't the culprit, a cheap generic PSU can fry your system at anytime it pleases.
m
0
l

Best solution

July 1, 2014 2:09:34 PM

Problem fixed. The fan on the PSU was dead somehow(would stop when the PSU started to get hot) and swapped with my old Diablotek fan. It's working great now.

So being new to these forums do threads close or just solutions found?
Share