whats a good tool to test stability of gpu

Solution
Furmark is good for testing a videocard, since it puts a pretty stressful load on the GPU. You're not looking at performance right? You just want to know if it can hold up under heavy use.

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
To caveat, different programs stress it in different ways. You'll see different loads on the card doing Unigine, Firemark, or Furmark. Your most thorough method of testing IS to run each of these and see what it does. Unigine can run just as long as Furmark. Firemark is more a benchmark. I found my graphical instability from OC in Unigine. Furmark didn't even make my card sweat.

Run unigine in the background while you're tweaking your OC, like my image below. You'll see the instability the moment it appears. When I hit my cap of GPU Core clock speed, everything crashed in a glorious 8 bit style Nintendo graphics failure. When I hit my memory core clock speed limit, I saw artifacting and texture load fails inside the app.
MAKE SURE YOU STEP YOUR OC INCREASES. Don't go up drastically. 10 vs 50. When you find your instability, just rope it back by 10mhz and then if you want to really push it, go up by 1 at a time. It can be painful, but it gets the most out of it without putting too much work on and damaging the card.
192364.bda46c679bd025fffe7a6179aaf6535b.1600.jpg
 

WINTERLORD

Distinguished
Sep 20, 2008
1,775
15
19,815
well now I'm pouzzled I ran an furmark and my rx 480 scored lower then a 470 at stock. I have a red devil factory overclocked score for the 480 was 2446

also noticed it didn't test very long only 60,000ms should it be running longer? or does the short test confirm stability
 

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
Different cards have different specs on them. It's largely effected by the core clock, but there's other cores that do different functions. Every program runs a card slightly different. Try out the different programs first to test stability across the board. Use Unigine in the background to tweak your settings, as it loops until you kill it (don't run the benchmark portion). Once you find the limit, stress it through other apps (firemark, furmark, 3dmark)

Then, bench it in 3Dmark and Firemark and COMPARE IT to similar cards and setups. See how it compares scorewise.
 

The_Staplergun

Estimable
Jan 30, 2017
1,395
0
2,960
Again, don't only rely in furmark. Leave it on until you think you're comfortable with the setting. It's up to you, as with furmark it does say use it at your own risk.

Follow the steps I outlined from testing and tweaking to benchmarking, it's a pretty all around check when you use all of that stuff. Only relying on one program won't fully see if you're good.