Why benchmarking varies so much using different software.

[PCO77]

There are obvious reasons why benchmarks May differ across different software. I'm wondering why or how it can at times be so drastic. For instance Novabench will give me an average fps around 70 and something like Tropics will bring it in around 150. Is it simply because some software target only the gpu? If that was the case then broader scoped software like Novabench would seem pretty unreliable. Maybe my understanding of benchmarking is wrong. I'm open to criticism of course or I wouldn't have posted.  I'm always open to learning.  Thank you. 

[Dujith]

Because they are completely different benchmarks build on different engines and thus will have a different result.

The older the benchmark the better the result to. So Tropics is older then Nova and thus will use less tech from newer generations.

 

But thats besides the point, as the correct way to use benchmarks is not to compare them to eachother.

But to provide a platform where different hardware will run the same test. So you can see if computer A is better/faster then computer B, not if benchmark A is better then benchmark b

 

 

[minibois]

Different benchmarks use different rendering techniques, textures, detail, etc. and thus will present you with different framerates.

Not every game will stress your system as much as the other one, same with benchmarks.

[PCO77]

Well that makes perfect sense. Many thanks. I've been into this for about 5 months now and I still have a lot to learn. 

 

[PCO77]

I finally got Timespy through Steam. Bye bye Novabench. http://www.3dmark.com/spy/9981976 

Edited January 1, 2020 by PCO77
Insert

[LogicalDrm]

13 hours ago, PCO77 said:

I finally got Timespy through Steam. Bye bye Novabench. http://www.3dmark.com/spy/9981976 

As the different benchmarks give so different scores, I suggest that you don't rely on one benchmark. As you may see from review videos etc. using multiple benchmarks to compare different hardware, or different OC settings, might give interesting results on what each setting or hardware is best on. Like difference between single core and multicore performance (Intel vs AMD in past 2 years). My personal benchmark collection has 5-8 tests, including 2 game engine tests.