Intel's 5GHz i9-9900KS is due to release next month

[looz]

Well you can take a look at Digital Foundry's CPU tests which almost exclusively test single player titles, and they've also pointed out that lowest FPS dips are CPU bound - in many titles even in 4K.


Also that it's preferable to not be CPU bound as that often leads to massive frame time inconsistency. Being GPU bound (or capped, preferably by in-engine means) is generally a lot smoother.

 

[NeverBackDown]

Yes the goal in development is to always be GPU bound. In certain cases maybe not, but for most you want to be GPU bound.

That's why DX12/Vulkan are CPU focused, to remove any and all bottlenecks by being more efficient, which in turn puts more focus on GPU performance and being a more important factor to game performance.

 

[grec]

Well you can take a look at Digital Foundry's CPU tests which almost exclusively test single player titles, and they've also pointed out that lowest FPS dips are CPU bound - in many titles even in 4K.

https://www.eurogamer.net/articles/digitalfoundry-2019-amd-ryzen-7-3700x-review

In Digital Foundry's 3700X review there's only 1 game at 4K with a more than 1fps difference in 1% mins against a 9900K and 5% mins and up are essentially identical, then it's usually only another 1 or 2 back to the 2700X for 1% mins on top, it's outside of margin of error but I'm not sure you could say it's a huge value incentive up there.

 

[looz]

I recommend watching the video linked in the article, specifically from this timestamp onwards:

https://youtu.be/SY2g9f7i5Js?t=449

It's more like .1% than 1%. But every thousandth frame still happens relatively often.

 

[gunbust3r]

Now I would have named it the i7+2 5775CX2++++++ but that's just me