New graphics card causes PC to freeze at random times

Relinkosed

Honorable
Aug 25, 2016
37
3
10,545
Recently I've bought a AMD Radeon HD 7790 card to put in my system that contains an AMD Athlon II x4 620, 4GB ram and an unknown PSU with "550W" (16 amps in the 12v rail), my former GPU was an HD 4670 1GB. Express uninstalled all old AMD drivers and installed the new Crimson Radeon latest drivers.

So I went and tested out LoL and CSGO, and they ran on a lower FPS then I thought they would, about 10 more fps then the last GPU at the same settings, but had no problem the PC didnt freeze nor did it stutter. And while browsing the internet the PC froze and I had to restart it. Thought it was a one off thing and didnt bother too much.

Next day comes and it happens again when on the internet and another time when extracting a huge file (~30gb) after a couple of hours (was a BSOD), after that I tried to seek help, and was told to do a Memtest and did it overnight and no problems at all, during the day did seperate stick tests and still no errors. The PC runs well on the old gpu, express uninstalled the new Crimson Radeon drivers and installed the old drivers suitable for the old gpu, and have been been using it again for 2 days and no problems at all.

Is the PSU causing freezes? It worked while in games and under load, but froze during regular PC usage
 
Solution
Okay fixed it. The SATA cables that were under the plastic part of the GPU were kinda bending it and it was a little out of place and so I got bent cables and it's been working fine for 4 days now.


That seems to make sense. Do you have the 6pin PCIe supplementary power cable firmly seated/plugged into the GFX card?

 

Relinkosed

Honorable
Aug 25, 2016
37
3
10,545


Yes, I made sure it was perfectly placed
 

Achint2000

Distinguished
Feb 10, 2013
692
2
19,165
First of all, "An unknown PSU" ??? A PSU can mess up everything if it's not good quality.

I BIOS OC'ed my Strix 970 on Corsair GS 700 and being one of the best brands in PSUs, in 6 months, it got burnt twice due to an overload on the overall +12v Rail wattage which I realized after it completely destroyed my motherboard.

You need to run serious benchmark tests and monitor them with a good tool. I'd say get Unigine heaven benchmark, set it to a good enough load and monitor voltages on GPU-z Monitor.

Then the new GPU has optimizations when it's not under load. I'm not much of an AMD person but on my Nvidia GPUs, when not in use, it switches to extremely low power states (P-States). My Strix 970's clocks go from 1468 MHz Core; 7700 MHz Mem; 1.212v to 135 MHz Core; 324 MHz Mem and 0.825v

Sometimes, when clocks are set way too low, it can cause crashes and errors. If it was Nvidia, I could list a simple set of commands to NvidiaInspector which would force a P-State on the GPU but I don't know anything about AMD.

Just get the GPU-z Monitor and while you're on the internet, see what clocks it drops down to. Atleast we'll know what the exact problem is. :)
 
You're using about 30W more power, which doesn't sound like a lot. But consider, you have a 95W CPU. Add an 85W-ish 7790, so there's 180W of demand on the 12V output right there. Your no name PSU is putting out 190W on the 12V, best case scenario. That's IF that PSU is even capable of providing its rated power over a prolonged period, rather than only for momentary peaks.
 

Relinkosed

Honorable
Aug 25, 2016
37
3
10,545


Took this picture of MSI Afterburner 3-4 days ago while I was internet browsing, was like this at times when playing a game as well

http://imgur.com/T3Amyo5

Looka odd to me, but what are your thoughts?
 

DSzymborski

Curmudgeon Pursuivant
Moderator


Honestly, it's not even worth worrying about a possible secondary issue when a whole array of issues can be caused by an unresolved primary one, a low-end junk PSU with a +12V rail that could barely power a sandwich.
 

Achint2000

Distinguished
Feb 10, 2013
692
2
19,165


Yes exactly, that's an extremely low P-State. There must be an Nvidia Inspector equivalent to AMD GPUs which can manually be used to force on P-States. If that's the clocks on gaming, it's terrible. 324 MHz is like using a GPU from 2004 or something. I'm a total noob when it comes to AMD.

Try MSI's K-Boost. As long as my search results on google, I didn't find anything as simple as Nvidia Inspector for AMD GPUs. Or find a simple small game on which the clocks stay high, alt+tab your way out of it and continue browsing the internet.

On my past 4 Nvidia GPUs, whenever I opened chrome or even skype, the P-States used to go to maximum. Even for a simple high frames GIF Animation. But this shouldn't happen. Low clocks are perfectly fine for internet, it shouldn't "CRASH". As for the crash, anyone would highly recommend to get a proper branded or certified PSU first and then check if the problem continues. In any case, a good PSU is worth it. No one wants their system burning down or getting destroyed.
 

Relinkosed

Honorable
Aug 25, 2016
37
3
10,545


I'll see what I can do, I know very little about changing the memory clocks or voltages at all, so it's going to be a bit harder :(
 

Relinkosed

Honorable
Aug 25, 2016
37
3
10,545
Okay fixed it. The SATA cables that were under the plastic part of the GPU were kinda bending it and it was a little out of place and so I got bent cables and it's been working fine for 4 days now.
 
Solution