My main point: Experienced a significant change in “smoothness” when changing a 6 core(5600x) to a 8 core(5800x), way less stuttering when all other factors remained constant. That’s only my subjectiv experience!
I am peeping around MSFS since early 2020, would describe my knowledge as advanced as nerdy overclocker who cannot take a flight without modifying something
I changed recently from a Ryzen 5600x (6c/12thr) to a Ryzen 5800x and the only change i made hardware (or software) was this processor. Same version of MSFS, same drivers… you get the idea.
The improvement in smoothness was amazing. 80-90% of stuttering (related to CPU, not the drops GPU related) are gone and beside the reload related stuttering from time to time, it’s a joy. All this in context and limitations of DX11, a great relief for now. DX12 will only improve that further…
I have always seen the heavy seven (7 threads taking the main work load) and all other cores with low loads. With the additional 4 threads of the 5800x i am experiencing that more often then not up to 10 threads are heavy and the other 6 taking smaller tasks (less then 20% util)
In the past i always recommended a 6 core for <4k, but with this result i had to alter my opinion. Sure, the 5800x’s load over all never exceeds 60% (low and fast over photogrammetry until reaching gpu limits) Maybe the additional threads don’t do much, yet it seems they are needed for smooth parallel processing. Maybe the switching between processes on systems with less then 16 cores is some cause of lag… Again, only my subjective experience i wanted to share!
PS: Hardware Unboxed recently compared which 8core is better for several games, MSFS was a part of that testing. The result showed that it doesn’t matter if you have a 10700k/11700k/5800x, the fps are within margins of error.
PS: I forget to mention 16 threads of the new generations of processors due to the higher IPC / Single-Core Performance. The older generations (even a 9900k are slowing you down again…)
I am aware, that are the limitations of DX11. This stutters will hopefully be solved with the rollout of DX12. In the envelope of DX11 i am experiencing a significant reduction of stutters even when landing on bigger airports… or over HQ photogrammetrie. It’s much better!
My argument was that 16 thread make it smoother. Better i had added 16 threads of the actual generation. What you are facing imo is that your 9900k has a lower IPC and therefore doesn’t work as much as the new generations.
9900k → CB20 single around 421pts
5800x → CB20 single 590 - 630 pts
So it helps that you have 16 threads but the problem is that your main thread is slower with the 9900k due too that dependency. Threads 2 - 16 have to wait for that main thread always a bit longer as it provides the “beat” aka fps.
Always check in the dev tool what is really going on. Sometimes the stutters aren’t related to your CPU or GPU. Slow Ram or wrong configured one can result in stutters, too or even reload stutters (if the ping is very low serverside)
That said, the stutters in your video look more like cpu stutters.
You state this advice is for “all resolutions”. How about when you are not CPU limited? I fly in VR (reverb G2) using a 3600x with 6900xt graphics. When benchmarking with the developer tool it shows I’m GPU limited. I’ve been wondering if I should pull the trigger for a 5600x or 5800x but couldn’t see a benefit given my benchmark results.
Your post makes me question that decision but you haven’t really given many details about your benchmarks before and after your CPU change, plus you don’t mention your GPU. Really I would have liked to see fps and frametime results for both the CPU and GPU before and after the change. The 5800x is not a cheap CPU. Any thoughts on my situation?
For example, I can get my frametimes down to around 20ms and lower for both the CPU and GPU, which virtually eliminates stuttering, by lowering the AI traffic settings and of course running with most graphics stuff at medium. But I would buy the 5800x if I could turn those up and still get minimal stuttering. I just don’t think it will make a difference for me.
Things are not black and white in gaming and in between Intel and AMD. Ryzen 3000 series performs slightly better in single thread over Intel 9th generation and gets run over by Intel in MSFS. You have to take into account architecture, latency, cache, etc.
Ryzen 5000 series fixed some of the problems, but it does not translate everything into fps in MSFS. Intel 10th and 11th generation usually gets right behind.
And the 5600X is more than enough for a DX11 game, even DX12. You could get the 5800X for rendering, streaming and such, but for who is on a budget, the 5600X is touted as the current best gaming CPU on the market. But depending on the price, the i7-10700K or the i5-11600K are good choices as well.
And stutter is subjective. What some people call borderline unplayable others call it “butter smooth”. I heard that a lot when MSFS was released due to the fact that general aviation simmers were used to low fps in X-Plane, while combat flight simulators were used to get 60+ frames in DCS and Il-2.
So from the i5-10600K to the 11900K and 5800X, you are good to go depending on your use. I would personally get the 10900K and disable HT - metal socket, real sensors, good overclockers, stable platform, etc.
i7 10700KF here but after two months of use I realize the sim does not take full advantage of it. May be with DX12.I believe ideal specs were 8c/16t.
I would say this if you have a powerful GPU and currently have a 6core 12thread cpu you should have no issues with games or this sim for the next few years.
The real boost for me would have been a GPU upgrade but the end of shortages are nowhere in sight.
One thing I agree with you the sim is smooth likely when I fly GA but once I hit my vram limit welcome back my stutter friend!
Overall mine is not a stutter free experience. Often limited by GPU.
Ryzen 3000 series does not perform better than Intel 9th gen with single threaded performance. It never did. Nearly every benchmarks shows the 9700k and 9900k beating the 3700x and 3800x in nearly every game .If you are referring to 5000series that would be more accurate. Intel was always faster in this area until the Ryzen 5K came out.
Not the same CPU, we discussed the k version, but you have a point here… 421 pts was the first result i found on google, i wondered because i taught it was around 20% difference. According to Techspot its around 509 pts CB20single.
Up to 20% difference, still makes a difference in fps…
I am running a 3060, so lower-midrange if you want, cause it’s the only card i could get my hands on for now… i am running 1440 resolutions in 2560/3440/5120 which lets me think i cover 1440p and near 4k(as pixel count). In my experience 1080p is a really dumb resolution to run a simulator thats created for 4k, and has similar demands on the cpu as 1440p… yet i am a sucker for graphic details so…
You’re running a high-end graphics card and are still GPU limited (not always i hope ), so you see that i am more often then not see the limitations of GPU with my RTX 3060. After all it isn’t really as simple as either you are gpu or cpu bound. Sure, the gpu sets at 100% load the frame limit to say 30 frames e.g. When at this 30 frames the CPU can keep up with 29fps, but every 30’th frame it can’t… there you have your stutter per second.
As i stated it is a subjectiv experience but it is for what’s my educated guess worth a pretty significant difference. With higher resolutions the cpu load increases even more…
Honestly, it’s Asobos recommendation to have 8 cores (even if they list weird/outdated models…) and if you are running highend reverb i would spend the extra money. Sure, DX12 will ease the situation to some extent, but as the sim evolves they are going to implementing features “beyond ultra” and therefore the demands on cpu will rise…
For me the upgrade was absolutely worth it. And in recent days the price of the 5800x dropped in germany to around 360€…
Thanks for the thoughts. I’m in Denmark so I buy mainly from Germany when buying online. You have me considering it. As I mentioned, I turn down the settings that push the CPU hard to ensure I’m balanced and avoid stutters. So I’m happy with what I’m getting but I’m always on the lookout to improve.
CPU lod increases with higher resolutions? That’s a new one to me unless this program defies the odds. What I understand is the GPU works much harder the higher res you go.I definitely hear my AIO fans rev up as I reduce to 1080p from 3440x1440.
I understand CPU plays a role with resolution too but for a higher resolution a better GPU is required. Its obvious.
Like my case my i7 10700K does nothing much on 3440x1440 while my low tier 2060super takes a good beating at this resolution and imo 4k was never worth it in the first place unless you have the hardware to deliver it. Great for screenshots though.
Its also not what I see on my Nitro 5 Laptop, with it’s modest i5-9300H @ 2.40Ghz, 32GB ram and very modest GTX 1650 4GB …
Admittedly I am running at 1080p, but i did not have to take out a 2nd Mortgage to get the Laptop, and running at 4K is not going to make me a better pilot.
You are applying known rules of gaming in general, an for that it’s true almost with every game.
Yet in a simulator like MSFS that relies heavily on AI generated stuff, not only buildings, CPU load in general is higher and with resolution also details that have to be displayed are increasing. (means cpu load increases)
Even if you have good quality photogrammetry, you always have many corrections to these buildings (height, form, depth color correction…) that are based on calculations by the AI/CPU “on the fly” in relation to your point of view, corrected by light, dependent on weather… you get the idea.
It’s more a question of what resolution can you drive with your gpu realistically @ first place and then choose a cpu that doesn’t hurt you/ ruin it with stutters on top. In your case the 2060 super isn’t gonna do it for the chosen resolution until you turn down the dials. So it’s underdelivering so much that the cpu has happy days… you can try to dial down the rendering scale to say 80% and you will see that the cpu now has suddenly much more to do…
), so you see that i am more often then not see the limitations of GPU with my RTX 3060. After all it isn’t really as simple as either you are gpu or cpu bound. Sure, the gpu sets at 100% load the frame limit to say 30 frames e.g. When at this 30 frames the CPU can keep up with 29fps, but every 30’th frame it can’t… there you have your stutter per second.