benchmark Which Cpu Benchmark Websites are Reliable? Why or Why Not?

[AMixOfGeekContent]

Hi. So I've been using these sites to compare Cpus and Gpus:

https://www.cpubenchmark.net/

https://www.videocardbenchmark.net/

They're just so convenient and nicely organized for me. Other sites I've tried, not so much.

Sometimes, if I'm watching a youtube video, they'll say this cpu is better than this one.

I check it on the above sites and yeah, the faster cpu has a bigger single thread score.

 

Which benchmark sites have reliable trustworthy data? What makes them reliable?

 

Thanks

[TylerD321]

I find it easier to look for youtube videos directly comparing the CPU or GPU with the rest of the build exactly the same. They tend to show gameplay with an FPS overlay. It's hard to directly compare across different generations otherwise in my opinion. 

 

Here is an example:

 

 

[emosun]

Those are actually pretty decent to get an "overall" idea of where the cpu roughly stacks up.

 

The reality is you have to check benchmarks on a per program basis if you want more accuracy.

 

Actually check benchmarks in the game or app you want to run. Otherwise, any general benchmarking  software is just taking its best guess as to what the load will be

[emosun]

1 minute ago, TylerD321 said:

I find it easier to look for youtube videos directly comparing the CPU or GPU with the rest of the build exactly the same.

One reason to kinda avoid youtube videos is they become outdated very quickly with game updates, driver updates , and os updates. They also tend to embelish the truth quite a bit as most arent from any reputable channels

[TylerD321]

2 minutes ago, emosun said:

One reason to kinda avoid youtube videos is they become outdated very quickly with game updates, driver updates , and os updates. They also tend to embelish the truth quite a bit as most arent from any reputable channels

To each their own. Of course if they are in the same video being directly compared you know they are using the same drivers or most up to date at the time. I just personally prefer comparing the general idea of which is "better" when the thing you are looking at is the only thing that is changed per system. The exact fps values and such are not really important to me, but I find it easy to quickly see how they stack up against each other. I wouldn't consider this method precise, just my preferred quick and dirty method.

[emosun]

1 minute ago, TylerD321 said:

To each their own. Of course if they are in the same video being directly compared you know they are using the same drivers or most up to date at the time. I just personally prefer comparing the general idea of which is "better" when the thing you are looking at is the only thing that is changed per system. The exact fps values and such are not really important to me, but I find it easy to quickly see how they stack up against each other. I wouldn't consider this method precise, just my preferred quick and dirty method.

Well again this method wouldnt be a good idea as the information can be manipulated and become outdated rather quickly. Which is why you typically never want to do that.

[AMixOfGeekContent]

I have another question: unlocked cpu's on cpubenchmark.net/. Are their results from the cpu being used regularly out of the box? Or from being overclocked to the max by someone who knows how to overclock? Cuz I don't know how to change voltage or whatever, and I don't want to do that cuz I'm afraid i'll break something.

[TylerD321]

I’d guess it’s a bit of both. The unlocked variants should still be clocked higher by default even if you don’t manually increase the overclock.

[brob]

1 hour ago, AMixOfGeekContent said:

I have another question: unlocked cpu's on cpubenchmark.net/. Are their results from the cpu being used regularly out of the box? Or from being overclocked to the max by someone who knows how to overclock? Cuz I don't know how to change voltage or whatever, and I don't want to do that cuz I'm afraid i'll break something.

 

The site reports aggregated data of a particular series of benchmarks. The methodology is outlined on the home page. As far as I know there is no taking into account differences in clocks between individual benchmark runs. But given the hundreds and usually thousands (that number is reported in the results) of tests aggregated the impact of a few overclocks is minor.

 

The site is useful for determining relative performance but it does not accurately represent the stock performance of a single CPU. It does however present an accurate approximation of the performance one might expect.

 

Ultimately system performance depends on so much more than the CPU.

 

 

[PDifolco]

I mostly use Cinebench rankings for CPUs (1C/All C), and TechPowerup database for GPU

Most "benchmark sites" use debatable methodology, where the whole of the user's PC setup impacts performance so you can't really tell

[AMixOfGeekContent]

1 hour ago, brob said:

 

The site reports aggregated data of a particular series of benchmarks. The methodology is outlined on the home page. As far as I know there is no taking into account differences in clocks between individual benchmark runs. But given the hundreds and usually thousands (that number is reported in the results) of tests aggregated the impact of a few overclocks is minor.

 

The site is useful for determining relative performance but it does not accurately represent the stock performance of a single CPU. It does however present an accurate approximation of the performance one might expect.

 

Ultimately system performance depends on so much more than the CPU.

 

 

Wow thanks. So I'm guessing only a very few people will actually dare to manually overclock their cpu.

[brob]

1 hour ago, AMixOfGeekContent said:

Wow thanks. So I'm guessing only a very few people will actually dare to manually overclock their cpu.

 

I don't think there is any hard data on the ratio of manual OC systems, auto OC, and stock. Most enthusiast motherboards of the last few generations implement a certain amount of OC through non-stock power settings by default. So many do run an overclocked system without knowing so.