Just bought and installed the RTX 4060 instead of GTX 1060.
Using Ryzen5 3600, 32GB I have to say that I hardly see an FPS imporovement. Yes, the overall feeling is a lot more smooth but FPS in most major airports is around 30-35. I tried to fly above NYC like in many settings videos on Youtube and getting ~45FPS.
Also, parts of the picture look a bit smeared with DLSS Balanced Anti Aliasing
It feels like I am not getting 100% of the GPU, cpu on 30%, 10GB/32GB are used, GPU on 35%, GPU mem almost full (8GB).
In dev mode I see under the fps “Limited by main thread”
My settings with the RTX 4060:
AA- Nvidia DLSS super resolution, balanced
Frame generation - ON
AMD Sharpening - 100
Resolution 1080P
Other settings - Mostly on High
Terrain LOD - 100
Objects LOD - 90
DirectX 12 (Beta)
Hardware and graphic acceleration - ON
Game Mode - ON
Nvidia settings - set high performance for MSFS and changed filtering to X16. Other than that all in default.
The 30% CPU value is meaningless. You need to look at the core load. If you see one core at 100%, that is the core running the main thread fully maxed out giving you a hard limit.
Also, running near the VRAM limit in DX12 will impact performance. Lower your texture resolution to get more headroom.
Interesting @PacyFire@MrFuzzy1337. I knew that FSX and P3D are very cpu demanding but I thought MSFS is better. I guess it is but sill not enough
This is how it looks:
Your screenshot shows that you have exceeded your VRAM and resulting memory paging is limiting your performance. Set Texture Resolution to “Medium” and you should see improved performance.
I have the 3080Ti, and I recommend
AA- DLSS Super Performance
Frame generation - off
AMD Sharpening - 0
Resolution 1080P
Other settings - Mostly on High
Terrain LOD - 150
Objects LOD - 150
DirectX 12 (Beta) - switch back to DX11
Hardware and graphic acceleration - ON
Game Mode - ON
Nvidia settings - set high performance for MSFS and changed filtering to X16. This is an urban myth. There is only one Anisotropic filtering. Set this in the game settings.
Set Texture Resolution to LOW.
Set Nvidia Reflex to On + Boost.
Turn down the shadows and reflections.
Turn off Ambient Occlusion.
Turn on Bloom.
Restore Nvidia Control Panel to defaults and set Max Framerate to 45fps.
I hope this helps.
As a matter of fact the filtering throuh Nvidia rather than from the game was giving me 3-4 fps boost with the previous GTX 1060 card. I checked it multiple times
The Ryzen 5 3600 is a 2019 mid-range CPU.
There has been a dramatic increase of performance since then, two big steps with Ryzen 5000, 7000 and the 3D Vcache versions that are absolutely the best CPUs to run MSFS.
If your motherboard can fit a 5800X3D, that would be the best upgrade you can do for little money.*
Also, as suggested, try to limit your settings in order not to fill your VRAM.
*EDIT: See this today’s post, what a coincidence
I tried to change to low but I see no real difference in FPS and also in look and feel. After restart I saw the difference. VRAM is down now but still FPS as you can see below in London Heathrow is around 30:
Not really. The problem is the CPU load, not GPU, so reducing GPU settings will have little to no effect.
For what it’s worth, London is a notorious resource hog, so you’re not doing horrible getting 30fps there with your current setup. Someone recommended upgrading to the 5800x3d, which is great advice if you can afford it. In the meantime, try reducing your terrain and object LOD’s a bit more, and your AI traffic (including roadway, water, and airport traffic). Those are the biggest CPU hogs in the sim. Don’t worry about your other graphics settings right now.
If you really want to try fine tuning, it wouldn’t be a bad idea to kill any unnecessary background programs. If you use Razer Chroma for RGB control, try switching to Signal RGB. I’ve found Chroma caused a lot of stuttering in the sim even though it doesn’t appear to be CPU heavy, and seen many others report the same.
Yeah, people still struggle to understand…
Main thread time = 31.6 ms per frame
Average GPU time = 10.7 ms per frame
It means that the GPU could deliver 90+ fps but it’s held back by the CPU that can deliver only about 30-32 fps.
This performance is totally normal for a Ryzen 5 3600 and there’s nothing wrong with it.
I have the same problem. I switched from i7 4770 to i5 13400F, GTX 1060 to 4060.
Mostly I have 40 to 60 fps, but at high complexity scenery, my fps go low as 30.
It is not a big problem, becasue it still smooth, compared to my old config. Before I had stutters, where landing was impossible.
Also limited by main threat, 2 cores are at 100%, 2 more are somwhere around 80% (not all the time), rest of the power cores are around 10-30%. So overall CPU usage is 30-50%. While GPU is around 80%. VRAM is not full, mostly it is around 80% usage.
I tried lowering terrain and objects LOD, but no change. Also tried DX12, CPU usage stays the same, 2 cores screams for life, other sitting there, doing almost nothing. I don’t get it, today, where we most CPUs have 6 cores or more, this sim isn’t optimised to use all of them or at least more that 2 cores.
Because from my 6 power cores and 12 threads, just 2+2 are used.
Just one parallel, I use this PC for ArchiCAD rendering, where render engine utilise all 12 threads, I see significant incrase in performance. Before I needed about 10 to 20 minutes for one render, now for same settings and same scene it takes only 2 to 3 minutes.
Gaming in general isn’t really capable of fully utilizing multiple cores because so much of it is sequential computing. What happens in one frame is dependent on the preceding frame and your inputs. Some have improved multi-core use, but more threads doesn’t always mean more gaming performance. With other processes like CAD rendering, all the information is already there, so the cores can work on more independent chunks of data at a time without needing to wait for variable inputs.
Maybe not the most technically accurate description, but hopefully good enough for laymen understanding.
