2700X - Upgrade Path

[CAT-THE-FIFTH]

OP just stick with what you have. Many enthusiasts like upgrading every 5 seconds, and with CPUs, it's a big difference if you are buying a £100 to £150 cpu every few years but not when you are spending £240+ all the time. If you are spending £250 on a Ryzen 5 5600X, a Ryzen 7 5800x at £340 is almost the same price per core and has higher clockspeeds.If that Ryzen 7 lasts you three years instead of two years with a Ryzen 5 5600X, the latter is not really cheaper. Plus the Ryzen 7 will have a higher resale value.

Plus in many games with RT, your RTX2070 is going to run out of steam quicker at 1440p than the CPU. It's all fine and dandy showing an RTX3090 with resolution scaling at 720p but from my own experience my RTX3060TI ran out of steam at 1440p with RT on well before the CPU.

On top of this in CB2077 I found the RT reflections the most noticeable effect anyway.

 

[Poneros]

The performance increase for the 5950X is huge vs the 5800X. What's going on in this test to account for the increase exactly? Not too familiar with the game but guess the combination of faster tuned memory, PBO and Curve Optimizer makes a significant difference.

It's definitely the tuning. You can see non-tuned performance w/ RT on and there's still a difference but not as large. Imo Cyberpunk performance is just presaging the future, because it's a very demanding game but absolutely uses any extra hardware you throw at it (and you can easily modify it to push further also). It's my main impetus for upgrading, as I can see my 6800K do all right in general but as soon as we're talking open world + raytracing it's just far from enough even for 60 fps.

 

[ljt]

Those are at 720p and 50% render scale though!

In a real world scenario the difference between those would be negligible

 

[Poneros]

Those are at 720p and 50% render scale though!

In a real world scenario the difference between those would be negligible

That's how you're meant to test CPUs. Think it through, those minimums are valid regardless because if you want 60 fps mins for a smooth experience then if a CPU can't do that at all then you will only find that out if you take out the GPU bottleneck - those minimums won't increase with resolution. So the only way to find out what minimums you can get out of your CPU is precisely a test like this. Otherwise you could upgrade your GPU thinking that that's what's holding you back and it wouldn't have been.

It's really no different than testing a car for how fast it is - you don't do it in a dune, you do it on a track.

 

[ljt]

I appreciate that is how you test absolute CPU performance, but its never going to be a typical use showcase.

Real world resolutions like the 1440p the OP uses would not show anywhere near the differences shown in that graph, especially with a rtx 2070.

For you to get those sort of differences at real world resolutions like 1440p and up would need an exceptionally powerful GPU that doesn't exist yet.

 

[Haz123]

The OP can upgrade to a Ryzen 5000 chip anytime he wants to with his mobo. Whether he would have a noticeably better gaming experience @1440p with a 5600x without changing his gpu is debatable.

 

[roadie]

I think there will be some AM4 chips with 3D stacked cache? I'm thinking it might be worth holding out for those so there isn't also the need for a new mobo and DDR5.