Question More FPS from upgrading from ryzen 2600 to 3600?

[rodrigueznickolaus]

Just wondering how much more FPS could be expected if I upgraded from a ryzen 2600 to a 3600 if all my other parts remain the same (X470 mobo, 1080ti, 16gb 3200 Mhz). I know it's too early to tell as the 3600 is not out yet but thought I'd check in with the experts here. Not interested in upgrading to a 2080ti, and all other cards are don't really hold a candle to the 1080ti (I guess the 2080 just matches it + ray tracing), I dunno. Additionally it also depends on the game but just looking for a ballpark estimate to get an idea. If it's too early, I'll just wait until reviews start coming in. Thanks!

 

[MrN1ce9uy]

We can hope it will increase a respectable amount, but until we see professional reviews released on July 7th, we simply don't know. Intel still claims they will remain the fastest in gaming performance.

I'm anxious to see if the Ryzen 5 3600 and all the other Ryzen 3000 can overclock to 4.9-5.0GHz.

 

[rodrigueznickolaus]

Oh wow, quick reply hahah, I thought as much but I figured someone might be able to theory craft some numbers lol

 

[MrN1ce9uy]

Well AMD claims ~15% IPC uplift, meaning that single-threaded performance (which is where AMD has been behind Intel) will be around 15% faster with Ryzen 3000 than Ryzen 2000. So, at best I would say 15% more frames per second, but it could actually be somewhere in between.

 

[rodrigueznickolaus]

Until reviewers get their hands on it, can estimate 0-15% more FPS then?

 

[InvalidError]

You get ~8% higher clock frequencies and 10-ish% higher IPC so you can expect 15-20% better FPS than equivalent 2000-series in situations of pure CPU bottleneck, less if something else (such as the GPU, drivers, APIs, inter-process communications, etc.) is hindering performance scaling. Zen 2 may be better, but not earth-shatteringly so for the same mainstream core count. Unlikely to be worth the trouble for most people who already own a 2000-series Ryzen CPU.

Performance gains will be larger on ThreadRipper and EPYC where centralized IO and memory on the IO die will reduce or eliminate NUMA-related issues.

 

[rodrigueznickolaus]

in situations of pure CPU bottleneck, less if something else (such as the GPU, drivers, APIs, inter-process communications, etc.)

ahhhh there's the catch. Thanks for your input.