I know the question has been debated a lot, but things are changing and that depends really much on the configuration, so I would be really happy to get the feedback from people with similar configuration as mine.
So I have an intel i5-9600k with an RTX 3080 and I play in VR with a HP Reverb G2. The performances are not as good as I would hope, they seem to be not as good as what some other people get from their 3080. Basically, I don’t manage to keep 30 FPS all the time, which I think is 100% mandatory in VR. What I do is that I avoid flying over large cities, as simple as that. And even then, I need to keep the detail level rather low (low LOD, most details on intermediate). I use OpenXR toolkit fixed foveated rendering and DLSS (and I am rather happy about it, unlike many people apaprently).
I will not update my GPU to an RTX 4090, which is just too costly and would require me to buy a new power supply and a new case.
So my question is, does it worth it to buy a new CPU, typically an intel i5-13400 or i5-13600 ? Can I expect an increase in FPS and less stutters in a very significant way ? Because really I would buy it only for msfs, for any other application/game my current CPU is good enough.
I can say that in my case upgrading my i5-9600K to a Ryzen 5800x3d gave my performance a boost. In your case the i5-9600K is not powerfull enough for the RTX3080 to deliver max performance. My RTX3060ti had also this problem and performs much better with the Ryzen 5800x3d. I’ am now no longer mean thread limited and fps is up and the sim runs much smoother.
Upgrading from an i9 9700K to an i9 12900K produced a significant performance uplift for me. Same GPU, a 3080 Ti. I recently went to a 13900K and got a further improvement. Honestly probably not enough to justify the cost of the CPU but I want to build another rig with the 12900K so it was worth it for me. Also, moving from DDR4-3200 to DDR5-6400 RAM seems to have made a difference for me. RAM can be a bottleneck in an otherwise fast system.
In general, yes. A faster CPU will give you some extra fps. I experienced it when upgrading from a Ryzen 3600x to Ryzen 5800x. I probably gained 5 fps increase.
But before you upgrade, I would open the fps counter in the developer menu first. It will tell you where you are bottlenecked. If it shows “limited by main thread” or “limited by rdrThread” then your CPU is the bottleneck. If it shows “limited by GPU” then a GPU upgrade would be better.
Actually, I had not tried that for quite some time, but with DLSS and OpenXR, it now shows “limited by main thread” almost all the time. I guess that’s it. How much will I gain, I don’t know but there is no doubt I will get something.
I have a RTX3080TI and went for a cpu/motherboard upgrade. Actually switched my i9-10900K for a new i5-13600KF. I clearly see that this combination gives me a much more balanced system. According to benchmarks my new i5 is performing 30-40% better due to new technology (PCIe4, bigger cache etc).
I recommend this cpu, and I think you will get a nice performance increase (40-60%) over your current i5-9600K:
I am currently pondering this very same question myself.
Regarding the suggestion to check the FPS view info in Developer Mode, this used to be quite a reliable method of determining whether you were CPU (mainthread) or GPU bound, however since SU11 I have found it to be quite unreliable and it quite often tells you that you are mainthread limited when you are actually not. This is evidenced by cross-referencing with GPU load in another utility (Windows task manager is OK for this but not with DX12). If your GPU load is at or near 100% then you are clearly GPU bound.
To illustrate this, in the following scenario (middle of nowhere at altitude in a simple aircraft) I should be GPU limited, as evidenced by my Windows Task Manager showing 100% GPU load constantly, yet MSFS Dev Mode tells me I am mainthread limited:
In that same scenario, if I disable turbo on my CPU and hence drop it from 4.8GHz to 3.6GHz, a 25% speed reduction, then there is a neglible resultant reduction in MSFS performance as you would expect in a GPU bound situation. If I am CPU bound, as MSFS Dev Mode is telling me, why am I not seeing a commensurate drop in mainthread time to the reduction in CPU speed? Something is clearly broken in MSFS Dev Mode since SU11.
Having said that, if you are running DLSS in VR, then it is most likely that you are actually CPU bound and your GPU load will be much lower than 100%. In this case, a CPU upgrade should be worth it.
Just to throw another spanner in the works, many of those that say they have experienced great MSFS performance gains by upgrading their CPU are talking about 2D experience. I have come across a few MSFS VR-specific posts that indicate that average performance with a CPU upgrade have been about the same, but with a significant improvement in 1% and 0.1% lows and hence overall smoothness is much better. Here’s a chart I saved of someones upgrade from a 5600X to a 5800X3D in MSFS VR and you can clearly see what I am saying.
I will add my 2 cents here. I have 9700k and 3080ti and i too have been tempted to upgrade my cpu to i7 13700k. However. There is clearly a great deal of problems in how MSFS is (ab)using the cpu. I’m pretty sure one of the biggest problems MSFS has in terms of performance is cpu related. I do not think you need the latest cpu (in theory), especially if you’re running VR at very high resolutions. Clearly, a great deal of optimization is needed and i personally have decided to wait a bit and see if anything moves in that direction.
Just to point it out…Games that have very poor cpu optimization hardly ever really manage to somehow turn that around. Best example of that is No man’s sky. Even through all the years of updating and optimizations, the cpu performance in that game is still extremely poor, so i guess it’s just not fixable. I’m not saying MSFS is the same. There are very clear benefits in upgrading to a cpu that has a good amount of L3 cache (for example). So nothing wrong with going down that road. But i would also point out that there is a high probability that 13th gen intel will be the last to use the 1700 socket. So it would make a lot of sense waiting for 14th gen before upgrading. Which is what i am going to do. I’ll build an entirely new system with 14th gen and a 5xxx nvidia gpu. By then the high speed (6000 mhz) rams should come down in price as well as that is important too. I feel going for 13th gen right now is throwing money out the window.
Most of us are running 3200-3666 mhz rams atm and that could bottleneck new gen cpus/gpus quite a bit (this is an unfounded assumption). I also personally feel no one should support Nvidia with its current pricing atm as anyone who actually is paying for their cards right now is doing a really bad favour to themselves in the long run. And also, pairing 13th with rtx 4090 for example, there is some concern the cpu might actually bottleneck the 4090 in certain games and resolutions. I highly doubt it could in 4k res, but at 1440p for example…in VR i doubt it even more though. But still… I will say the current hardware is not very optimal and especially if you consider prices, i’d say wait.
OK, thanks for all your answers, I have bought a new proc and motherboard, Ryzen 7 5800x3d. I still have my RTX 3080.
The frame rate is definitely much better now. How much I can’t say, but I know manage to get 30 FPS constantly (retroprojection locked), with reasonably high details/resolution, without having to go through the painfull process of updating the detail level depending on where I fly, reducing it each time it starts stuttering, which is awfully breaking the immersion. I still use DLSS/Quality, which I like. I also put volumetric cloud on High or Ultra, which is very costly but worths it in my opinion.
There are some things I still don’t really understand, maybe I’ll open a new thread. Right now, my GPU is not fully loaded, actually far from it, even when I am GPU bottlenecked.
I really feel like in this game things are more complicated and there are no clear distinction CPU bottelneck / GPU bottelneck, things are not that simple. I try to understand the output of OpenXR Toolkit (app CPU, app GPU), but they don’t seem to be as obvious to interpret as it seems. I’ll try to dig a little bit further, and maybe open a dedicated thread.
I agree, I have a 3080 went from a i7-9700KF to an i5-13600KF with 32GB DDR5 and didn’t notice a huge difference. It was mostly to improve performance at large airports where I was CPU limited, but I still have to turn off all traffic to get my FPS above 50FPS on the ground.
Yes it’s really complicated, and there’s no silver bullet. You have so many different scenarios in the sim, some are very heavy on the CPU, others very heavy on the GPU. And so many settings you can change. I’ve seen youtube videos where a i9-12900K struggles to get above 30FPS on the NY landing challenge, or landing at O’Hare airport, I’m pretty sure they had LoD at 200 and they must have had airport traffic on, but still, you’d normally expect a CPU like that to generate 60FPS at least.