Do overclocking software such as MSI / NVInspector change anything when overclocking?

daynwsowulf

Reputable
Feb 25, 2015
39
1
4,540
Hey Tom's hardware experts,

I've been trying to over clock my GTX 970 and I made it to the 1500mhz. So far so good, but I am in doubt as if is there any difference between these over clocking software.

I first tried with MSI Afterburner, all I've done is change the clock speed to +260, and then I was getting 1500mhz boost when playing, but it would crash the driver at random times, usually after 10/15 minutes of game play.

Yesterday I tried with NVidia Inspector and I did the same thing, +260mhz, nothing more, just the core clock speed, and surprisingly, no crashes, played games all day with 1500mhz and no crashes at all.

So I came up with this question, could over clocking utilities operate different from one another?
 
Solution
The issue with the driver crashing was the Overclock was to high. This is a standard and very common response to an unstable overclock.

Can they operate differently? yes an for the most part no. Confused? Yes in the aspect they can be coded to function slightly different but for the most part they all do the same thing which is modify the software so that the clocks run at a different speed than stock so in this aspect the answer is No. I recommend using OCCT and testing the GPU with error checking enabled. this will actually tell you if the overclock is unstable. If it detects errors then it is unstable.
Different games and scenarios will provide different results when OCing. I would recommend downloading something like the Heaven benchmark and running it every time you change the core or memory clock speed. It takes a while but it will give you the best results. If it crashes take the clock speed down a little and try again. Every card is different and I doubt that the program you use will affect how much you can OC unless the software is absolute crap (MSI Afterburner is really good and what I use). OCing can be tedious sometimes because it might work fine for a week with the OC then one day crash in a game and you have to tweak it a little.

I would adjust the core clock speed until you get it just right then start playing with the memory clock speed. It works best this way in my experience.
 
The issue with the driver crashing was the Overclock was to high. This is a standard and very common response to an unstable overclock.

Can they operate differently? yes an for the most part no. Confused? Yes in the aspect they can be coded to function slightly different but for the most part they all do the same thing which is modify the software so that the clocks run at a different speed than stock so in this aspect the answer is No. I recommend using OCCT and testing the GPU with error checking enabled. this will actually tell you if the overclock is unstable. If it detects errors then it is unstable.
 
Solution


This is not the way to go about overclocking a GPU. while you can use Heaven for telling if the overclock has helped any it will not detect the errors that appear so fast the eye can not see them. this is where OCCT's GPU test comes in. It actually has an Error Checking option that will check for errors you can not see and tell you if the OC is stable. Just FYI always adjust 1 clock at a time never 2. this will help knowing which setting is bad and which one is good.
 
Another thing to consider adjusting when overclocking is the Power Limit%. I believe for the 970 this can be adjusted to 125%. This will allow Turbo Boost to use a higher power target before throttling your core clock.

I think it might have been just coincidence that it never crashed while using NVInspector. Both NVInspector and Afterburner write to the same registers in the GPU to control the core clock.
 


Didn't know about that software thanks for the heads up!



Isn't that what I said?
 


Yes but in my experience with overclocking, and not just on GPU's mind you, I have found that adjusting one clock and finding the max then setting that back to stock then adjusting the next to max then testing them both or all together yields the best and highest combined overclock. Besides that part of the statement was not meant towards you but in general.
 
I have to agree, you should find the highest core clock first. Then return it back to default before find the highest stable memory clock. Typically with GPU's the core clock affects performance the most, so you always give preference to core clock over memory clock.