Micro- Stutters - A possible resolution for some.

[Lord Farringdon 498]

 

Hi Guys,

(Note to Mods: This post is probably in the wrong place but a lot of guys with micro-stutter issues may not see it tucked away in other less visited forums, so maybe a couple of days grace before shifting it?)  

Some of you will know that I have been carping on about micro-stutters for some months now with no solution to hand. Others have also experienced them and have given up trying to find a solution. I may have discovered at least one solution to these micros-stutters and thought this might help others who have thrown their hands in the air regarding this issue. Everyone sees stutters differently and often describe them differently too which makes it quite difficult to determine whether the solution is valid in your particular situation or not.  Additionally, everybody's systems are more or less unique and the combinations means something that fixes my problems for example, might not fix someone else's. So with those caveats, let's start by saying this is not a Microsoft, Asobo, MSFS or poorly optimised software addon problem. And, it's not about settings in MSFS. All of those types of stutters, are just like the shakes you get when you try to lift too many weights!! Improve your strength (better processing power) or dial down your workload (lower settings) and the problem resolves itself. But pure hardware issues may not respond to those sorts of measures and so we have to look closer at our components including BIOS, CPU loading, GPU and memory to name a few. Don't worry, I'm not going to explain that in detail here!!   

Firstly some definitions: 

Stutters: Pause's, stutters or "buffer", which are seen in games, streaming, video's etc is caused by the speed of your Internet connection and / or the processing power of your computer. Effectively, the graphics card can't display something it doesn't have and so it begins to skip frames to keep movement going or it simply pauses. 
 
Micro-stutters:  As the name suggests they are very short duration (micro) drops in FPS which causes the image to very briefly stop or slow down. If you are panning around it is just a momentary glitch. If it occurs only very occasionally, it's not really noticeable. But, when happening randomly, every few seconds, these glitches become a real immersion killer. In my case, they were not just happening while flying or when panning around on the ground but even while waiting at the World Map to load a flight!! Of course if there is no movement, you cant see a micro-stutter (because the image is not changing and doesn't need to be updated anyway)...but it's still there in the Development Mode FPS counter. You can see them in this video as the long spikes and milli second drops in frames of up to 7 or 8 FPS in some cases. This video was taken while sitting at the World Map screen. 

 

So I did the things we all do in these circumstances and that is to isolate everything and reduce all the other possible influences on FPS. Disconnect internet, remove everything from community folder, lower or turn off all MSFS settings, especially live weather, AI and multiplayer (internet disconnected anyway), and then for good measure ensure the video card had the latest driver,  BIOS was updated and system memory was tested. I then started MSFS and noticed Home screen micro-stutter spiking occurring in the FPS counter without even loading a flight!! I opened Task Manager to compare the 16 logical CPU performances with the spiking and while there were some similarities there did not appear to be any direct correlation. 

Things changed however when I looked at the drive performance in Task Manager. The C drive was spiking almost exactly at the same time as the FPS counter in MSFS. The spikes were not overly expensive on drive capacity maybe only by 25% and I thought this may be caused by a process going on in the background. So, I killed all background and startup processes, including Windows Defender but the drive still continued to spike.

 

My C:Drive is a Crucial NVMe P2 1TB, as in this image below, so I looked up the Crucial site and downloaded an update for the drive.

 

This helped reduce the micro-stutters but did not eliminate them. Still, that was a real positive and suggested I was looking in the right place! Then I found that the software which I was required to use to update the Crucial drive (Crucial Storage Executive), contained a product called 'Momentum Cache'. I won't go into the detail of what that is but when enabled it is designed to use a bit of system memory to help the NVMe with caching and buffering. This seems to overcome an otherwise less than stellar performance in the Crucial P2 NVMe's. 
 

The result of enabling 'Momentum Cache'  is that my NVMe C:Drive is now handing data to the CPU much quicker, rather than spiking which suggested it was doing something other than sending data to the processor! Taken from the World Map screen again, here is the video of the outcome. No more micro-stutters and an improved main thread performance!  

 

The whole science behind SSD NVMe's is quite complex especially around things like 'Dramless QLC' (that's all you need to know about that). Yet, the NVMe (or any drive) is such a critical part of processing that if it is below standard (which some of these Crucial ones seem to be), then bottleneck micro-stutters start to show. The term 'under buffering' is probably more appropriate since it related to data and applications including audio and video so it might not be just our MSFS flight visuals that are affected by a hardware related micro-stutter. 

So if you have an NVMe, especially if it is a Crucial P2 (Crucial are usually the cheapest and a lot of computers and laptops come with it pre-installed to keep costs down) and you have micro-stutters, then this may just be the cause. I have no idea if other brand NVMe's or later model Crucial NVMe's might also cause these issues but it is a thing to research if you have other types installed on your computer and are suffering micro-stutters. 

So there you have it. Micro-stutters are a persistent issue for some of us and difficult to resolve. That could be because some of the origins of micro-stutter are so obscure. Anyway, I hope this helps to uncover at least one aspect of the problem. 
 

Cheers

 

Terry

[turbomax 3,290]

Good Lord 😊, what a great detective work, that's exactly the correct procedure to identify and eliminate such hard to detect performance issues. and another reason why I never buy the cheapest, but usually the more  expensive ! brands, and when it comes to NVMe, I have never purchased anything but Samsung. and while other manufacturers are already shipping 4 TB, I have been holding out for their new 4 TB 990 PRO NVMe M.2 SSD, PCIe 4.0, 7.450 MB/s read performance out on October 2nd. enough to hold MSFS 2020 AND 2024. 😊

btw: "Micro-stutters are a persistent issue for some of us and difficult to resolve. That could be because some of the origins of micro-stutter are so obscure."

very true.

sufficient cooling is also important for the latest high speed NVMe, otherwise you may encounter automatic throttling without you knowing it.

Edited September 25, 2023 by turbomax

[Bluescaster 21]

Hi Terry,

Many thanks for sharing this with the Computer 'Numpties' like myself. A very interesting Post that, quite clearly, you put a lot of thought and effort into (even if much of it was above me;-) ).   I am always grateful for the generosity of fellow 'Simmers' sharing their (in this case, quite extensive) knowledge.

Thanks again.

Kind Regards, 

Blue

[Vensaj 6]

Please be aware of risk of data loss with this enabled... 

[Mace 1,932]

That's an interesting fix and I'm glad you posted it.   I've always considered Crucial drives to be on the upper tier of makers.  I researched a Crucial P5 Plus this winter/spring and the results of that research mean that it graces my new 7800XD sim rig as primary storage.

Samsung were having some issues this winter/spring but they used to be my go-to.  They may be yet again.  I do not believe in writing off any tech maker for all-time because manufacturing changes all the time, components like IC's can change etc. etc..  Anyway it's good information re: the P2 and if I ever have that problem I'll remember your findings.