MSFS 2020 Support Update 03.09.2020
22. August 2020 at 18:05 #1905
I believe that the release version of MFS has performance issues in its SimConnect code. 3rd party applications which use SimConnect and worked fine with the alpha and beta releases of MFS now produce stutters in the release version.
I also believe that the software running our yoke is either buggy or reckless and produces an incredible amount of SimConnect requests. Last year, I used FSUIPC to log the SimConnect events by CLS2Sim. After just 10 minutes of logging, FSUIPC created a logfile with 600,000 lines of output. I am not sure how many lines of output each SimConnect request results in, but it might well be that CLS2Sim (in the version that I tested) produced up to a 1000 SimConnect requests PER SECOND.
Now combine those two observations and it seems very possible that once CLS2Sim connects to MFS, the sim will spend almost all of it’s time processing SimConnect events… thus it hangs.
One could argue that Asobo “just needs to fix” the performance issues, but one might also say that perhaps Brunner could do its part as well to reduce the load on SimConnect by checking their code for the reason of the very high amount of SimConnect requests.
I did not inspect CLS2Sim’s behaviour for fun, but because I had other issues, which were, in my opinion, caused by overwhelming SimConnect on my machine.
Caveat: I have not used the alpha or beta versions of MFS myself. I have not used my yoke for months (out of frustration with certain issues). I’m back to flying with my Warthog, even though it doesn’t provide the kind of immersion the yoke might provide.
Perhaps someone who’s a bit tech-savvy can look into this, check whether the number of SimConnect requests from CLS2Sim is still as high as it was when I looked at it. Not sure if FSUIPC7 can log SimConnect events yet, but I guess you could just use your P3D install, the behaviour of CLS2Sim is probably the same.
If you can verify the “up to a thousand SimConnect events per minute”, you might ask Brunner about it. I think I have lost the goodwill of the Brunner support people. I raised this issue with Brunner support in October 2019 and never received a response.23. August 2020 at 11:40 #1906
Correction of the previous post: In the last paragraph, I wrote “events per minute”. It should say “events per second”.
Just to clarify: the rate of SimConnect events caused by CLS2Sim and logged by FSUIPC in my experiment was 60,000 events/min = 1000 events/s.
Even considering that the yoke is supposed to act and react without the user perceiving any latency, this feels excessive considering the limited number of datarefs CLS2Sim actually needs to read and update at a high rate. And since, on the sim’s side, the SimConnect API appears to be part of the “main code” that needs to run in the main thread of the application (ie. as part of the code that runs on the infamous first core which is always at 100% usage for most simmers), this appears to have a performance impact which is especially noticeable (e.g. as “micro stuttering”) when other SimConnect applications enter the equation.24. August 2020 at 0:23 #1907
During my experiment (in post #1898), CLS2Sim was reporting the connection as 90 events per second.24. August 2020 at 10:20 #1908
My problem is more probably caused by FSUIPC7 (Beta) rather than by CLS2sim itself.
I Don’t know if it is due to CLS2Sim + FSUIPC7 or only an FSUIPC7 bug (API MS/ASOBO?)…
Anyway, I hope you’ll release soon a version able to get force feedback with ALL MFS2020 aircrafts… and that solve the well known old problem already reported (TFDi B717 / MilViz KA350i).
Didier Pilois24. August 2020 at 10:49 #1909
CLS2Sim reads around 65 SimConnect variables and writes about 30 at varying frequencies.
All in all, 1000 updates a second is actually a somewhat accurate measurement.
Ist it much? Yes, but P3D and MSFS X had no problem with that.
We could lower that, but that would worsen some features (ground vibrations and turbulence for example).
Now when I searched for the issue with the launch version, I lowered the update rate to 1 update per second and also deactivated each API call separately.
I still had heavy stutter. So there definitely is an issue with the API.
As mentioned previously, that alpha build hat no performance issue with the heavy update frequency,
so they must have changed something.
Apologies to daddyboon and aurel42.
I’m sorry for leaving you in the rain.
Im bumping up your tickets to the front.
Diego24. August 2020 at 11:03 #1910
Thank You Diego 😉
One more line to add that “Massive FPS drop” using FSUIPC7 beta is confirmed (https://forum.simflight.com/topic/90203-massive-fps-drop/page/3/).24. August 2020 at 11:08 #1911
I can only agree with Diego’s words.
It is definitely a Microsoft / Asobo problem with SimConnect (which there was also a short time in the alpha version).
I can also confirm that FSX, P3D and MSFS2020 Alpha with 1000 updates per second by CLS2SIM never had a problem – please do not make any changes at this point!
Greetings Michael24. August 2020 at 12:00 #191225. August 2020 at 3:36 #1914
Hey Diego, I searched the setup manual and cant find using USB mode. Do you have a link to the proper documentation? Thanks!28. August 2020 at 7:38 #1918
Fix is comming soon
Stefan Brunner28. August 2020 at 11:07 #1919
Thanks for the update Stefan, this is great news!2. September 2020 at 20:33 #1920
Patch is out, i am not able to test it but it has been reported as fixed.
read also here
more infos tomorrow
Stefan Brunner2. September 2020 at 20:42 #1921
That’s great news!3. September 2020 at 2:15 #1922
Just did a very quick test with the Cessna 172 and was able to fly a quick circuit.
One thing I noticed though was how turning the yoke left/right and comparing the position of the physical yoke with the yoke in the v/c in the Cessna, when I moved the physical yoke about 80 degrees, that would only be like 45 degrees in the Cessna in MSFS. Then when I moved the last 10 degrees of the physical yoke, the yoke in the Cessna moved very quickly all the way from like 45 degrees to 90 degrees.
Couple of other things. It felt like the sensitivity was very high. Also, the forces required to move the yoke along both the pitch and roll axes felt very soft. I’m sure all of this can of course be fine-tuned to match various aircraft etc.
Would be awesome if you could come up with some recommendations and/or a guide how to setup your Brunner yoke to work in the best possible way with MSFS once you had a chance to try it out yourself. That would be highly appreciated!3. September 2020 at 2:44 #1923
Works fine for me, I’m getting feedback based on airspeed over the controls as expected.
@webmaximus – maybe you need a different profile or tweaks to your profile for the Cessna 172? I’m using my X-Plane 11 172 profile (adapted from Stefans) and I’m not seeing the same behavior you reported.
One question for Brunner… is CLS2Sim still supposed to connect via USB HID/vJoy, or should the yoke/pedals show as native devices in the sim now? I’m still seeing USB HID/vJoy based setup, even after selecting the “flight simulator x” connection method in the CLS2Sim settings.
You must be logged in to reply to this topic.