Most accurate way to compare CPU's? single/multi core

[Brumboy]

So I downloaded Cinebench yesterday to run some benchmarks to see where my CPU currently compares to all the new CPU's. Is Cinebench a good way to compare single core and multi core performance between CPU's, specifically for gaming purposes? Pretty much every review out there contains a Cinebench benchmark test.

For example here are some Cinebench scores from TechSpot and TechPowerUp for a range of CPU's, with my CPU (4690k at the bottom):

10900k - single = 540 = 33% faster than 4690k
10900k - multi = 6400 = 407% faster than 4690k

9900k - single = 510 = 25% faster than 4690k
9900k - multi = 4870 = 310% faster than 4690k

3700x - single = 508 = 25% faster than 4690k
3700x - multi = 4900 = 312% faster than 4690k

3600 - single = 488 = 20% faster than 4690k
3600 - multi = 3740 = 238% faster than 4690k

4690k (4.2) - single = 405
4690k (4.2) - multi = 1570

Is it a fairly accurate way to compare these results and use them for a comparison for choosing a CPU for gaming?

 

[Brumboy]

Well it gives you the numbers for single and multi threaded increases. In terms of choosing a gaming CPU I don't think you can go wrong with any of them, even the 3600 has 3x as many threads as your 4690k and from what I remember with my 2500k on my son's machine, thats what made the biggest difference.

Rocking a 3700X now and not looked back

It was more of a general question, because comparing fps benchmarks isn't always accurate since the test setups are completely different to my own. But using synthetic benchmarks I assume would be a fairer comparison.

Yeah I've been looking at an upgrade for months but thought I would benchmark my own CPU and compare it to what's out there.

Nice, was close to getting a 3700x!

 

[Brumboy]

It's not the whole picture but it's a reasonably good indicator.

Would you say it should have a major impact when buying a new CPU for gaming?

 

[Brumboy]

Ah right, is there a better way to compare gaming performance with a benchmark then?

 

[Brumboy]

Not without testing a load of games and checking your framerate. Steve at Gamers Nexus doesn't do testing marathons for the fun of it. It really is the best way.

Are you actually unhappy with the framerate of your current CPU?

True I will look at his reviews again.

I play esports games so nothing demanding in terms of graphics but there are stutters and the min fps is low playing at 144hz. I literally turn every graphical setting to low to maintain 144hz in all situations. As in I'm pushing my CPU to the limit at 1080p.

Apologies, yes its as good as any I guess, I wouldn't get too hung up on single core these days as long as its in the ball park, threads are far more important especially with the arrival of the new consoles, unless you are chasing the 1080p dream, then I guess its intel

No problem! I really would avoid Intel as the upgrade paths are so limited. My CPU budget is around £300 which would get me a 10600k or a 4700x in the future. Obviously both of these CPU's have a lot more threads than my little 4690k!

I had a 4690K running at 4.6Ghz, compared with the 3600 i have now the difference is huge.

What GPU do you have?

I have a Zotac 2080 Extreme Core, play at 1080p/144hz. You say the difference is huge can you give me some examples in terms of fps numbers please?

When testing CPU benchmarks, you will typically find that the benchmark is done with top end GPU & RAM, and the CPUs they're comparing.

E.g. if you're looking at a benchmark for 10600K, 10700K, 3600X, 3700X then they will usually be paired with a 2080 Ti.

They do this because they're trying to measure the maximum potential differences, by reducing any potential bottlenecks from the GPU and RAM and putting it onto the CPU. In addition, you will notice that some people will review CPUs at 1080p medium or low settings - another way of pushing the load onto the CPU, helping the reviewer to measure maximum differences.

While this does indeed provide the best insight into how good the CPU actually is, it isn't always representative. Why? Well, not everyone would pair a 3600X with a 2080 Ti - these differences will be less pronounced as you move to a less powerful GPU, where the GPU again becomes more of a bottleneck.

As a result, you might find that the differences between your 4670K, 3600X and a 9900K at 1080p max settings with a 2060, for example, are much lower (4-5%).

I don't know of any benchmarks that use midrange GPUs for CPU gaming tests, but it might be worth checking out.

Thank you for the informative reply. I totally understand.

Which reviewers tend to test at 1080p medium or low settings?

See that is what I am worried about. Upgrading everything then finding in the real world not much difference between both setups.

I am looking at getting a 280hz monitor though, very tempted. If I get one then I would defo need a top CPU!