FPS Degradation on Long Flight
-
I wish I could say that removing toolbar pushback fixed things, but I can confirm that the issue is still present without it. It happens after Starship has been loaded up and operated continuously for around 8 hours. Just did Gander to Keflavik, then Keflavik to London. Each leg about 4.5hrs, with an hour on the ground in Keflavik. Right around 3 hours into the second leg, performance began dropping steadily. Maybe a memory leak somewhere?
-
I never used Toolbar Pushback to begin with and still have degrading FPS and/or MainThread lag spikes.
-
Okay, anyone who wants to test this for me, here is the first thing that I would try. It's extremely simple.
Locate
...\bksq-aircraft-starship\html_ui\Pages\VCockpit\Instruments\NavSystems\bksq-aircraft-starship\MFDmapConfig.jsonChange
preventLabelOverlaptofalse. That's it.There is a chance that you will have to drag the aircraft's layout.json on top of the MSFSLayoutGenerator afterwards, but only in MSFS 2024.
If that doesn't work...
I have already heard confirmation that removing both ND's and the MFD from each panel.cfg eliminates the problem, but it would be nice to get some additional confirmations that this is indeed where the problem lies.
To do this, you can either remove or comment out the three [VCockpit] entries in each panel.cfg (or in only one aircraft, if you're feeling confident of which is which) that include "MFD", or "ND". No renumbering in the file is required. Comments can be made by beginning each line with
//or;.After that, I'm afraid that it will take some volunteers to solve this one again. If any of you have already written to Just Flight about the waypoint deletion CTD, and also experience this problem, then let me know.
Much appreciation again. As you might imagine, these bugs that only affect some number of users, and none of my beta testers can be a real pain to track down, so I am very thankful for your willingness to help.
-
That will simply remove the ND and MFD from the aircraft completely (you will get white screens in the cockpit, which, if I'm not being too cheeky, is a great excuse to practice your reversionary modes). Like I said, I already had confirmation that the problem lies with the ND or MFD, but I would like to have more than one data point.
The first, extremely simple test would rule out one thing that I changed/created in the MFD map, but not the ND. If it doesn't fix the problem, then I will know where else to look.
Of course, it makes little sense that this is another problem in Starship that only affects some users, but that's software, I suppose.
-
Changing preventLabelOverlap to false (and regenerating the Layout.json) might have imprved things a bit but it might be a fluke. I did only one 2 hour hop with it. The performance issue is certainly not resolved.
Do I change that setting back to true before testing flying without ND and MFD?
Another datapoint in terms of platform:
Windows 11, latest version. 13900K, 5090, 64GB DDR5 RAM.
I already lowered resolution and settings. GPU is at 30% load. Just to make sure it's not somehow graphics related. -
Okay, anyone who wants to test this for me, here is the first thing that I would try. It's extremely simple.
Locate
...\bksq-aircraft-starship\html_ui\Pages\VCockpit\Instruments\NavSystems\bksq-aircraft-starship\MFDmapConfig.jsonChange
preventLabelOverlaptofalse. That's it.There is a chance that you will have to drag the aircraft's layout.json on top of the MSFSLayoutGenerator afterwards, but only in MSFS 2024.
If that doesn't work...
I have already heard confirmation that removing both ND's and the MFD from each panel.cfg eliminates the problem, but it would be nice to get some additional confirmations that this is indeed where the problem lies.
To do this, you can either remove or comment out the three [VCockpit] entries in each panel.cfg (or in only one aircraft, if you're feeling confident of which is which) that include "MFD", or "ND". No renumbering in the file is required. Comments can be made by beginning each line with
//or;.After that, I'm afraid that it will take some volunteers to solve this one again. If any of you have already written to Just Flight about the waypoint deletion CTD, and also experience this problem, then let me know.
Much appreciation again. As you might imagine, these bugs that only affect some number of users, and none of my beta testers can be a real pain to track down, so I am very thankful for your willingness to help.
@Black-Square said in FPS Degradation on Long Flight:
Change
preventLabelOverlaptofalse. That's it.Happy to report that I repeated the same 3.5 hour flight that previously caused lag spikes, and this time there were no MainThread issues I would attribute to the Starship rather than MSFS 2024 SU2 after applying the change. I'm cautiously optimistic for now. The sim is still running, and maybe I'll manage to do the return leg later, despite the heatwave sweeping through Europe right now. Win 10, 5800X3D, 4070 TI Super, 64GB ram
-
Thank you both so much for testing! Please, please continue testing with that flag set to off. I wasted about a week writing better code than what Asobo had there, but ultimately decided that it was too heavy on performance, so I rolled back to Asobo's code. Apparently theirs wasn't much better! Ironically, I discovered that Starship didn't even have the processing power to implement that feature in the 80's, so removing it is the most realistic thing to do anyway! I don't know if that's what I'm going to do yet, but please, as many people posting in this thread as possible, please test this potential solution!
-
Thank you both so much for testing! Please, please continue testing with that flag set to off. I wasted about a week writing better code than what Asobo had there, but ultimately decided that it was too heavy on performance, so I rolled back to Asobo's code. Apparently theirs wasn't much better! Ironically, I discovered that Starship didn't even have the processing power to implement that feature in the 80's, so removing it is the most realistic thing to do anyway! I don't know if that's what I'm going to do yet, but please, as many people posting in this thread as possible, please test this potential solution!
@Black-Square Okay so I once again don't want to speak too soon, but here's what I've got.
Changing preventLabelOverlap to false didn't do anything to solve my issue. So I finally dove into the BIOS and enabled C-States. That simple change has given me a MASSIVE increase in overall sim performance, and I'm hoping it's also the solution to the gradual decrease in performance after having Starship loaded for many many hours.
Im also an AMD user, and apparently AMD sees the biggest benefit from it.
Will advise, but the initial impression of the overall performance improvement is nothing short of incredible.
-
I'm beyond excited to report that enabling C-States on my AMD motherboard has completely eliminated this issue for me. Did a 4 hour leg, followed by a quick turn to a subsequent 3 hour leg, and its still running silky buttery smooth.
@Black-Square And given that changing the labelOverlap value technically makes the MFD more realistic, I'll be keeping it that way
-
I'm beyond excited to report that enabling C-States on my AMD motherboard has completely eliminated this issue for me. Did a 4 hour leg, followed by a quick turn to a subsequent 3 hour leg, and its still running silky buttery smooth.
@Black-Square And given that changing the labelOverlap value technically makes the MFD more realistic, I'll be keeping it that way
@SinkRate said in FPS Degradation on Long Flight:
given that changing the labelOverlap value technically makes the MFD more realistic
Enjoy suffering with the power of a 5 MHz 8087, haha.
Thanks for the report on the C-States. I'm glad this is fixing the issue for so many people, but it's still concerning. If case you can't tell, these issues that only present on some hardware with some BIOS settings are really a developer's worse nightmare. Please spread the word about this fix, but I am also happy to explore deeper with anyone else who can spare the time.
-
Just for clarity and tracking, this is my timeline of events:
Flight 1 (1.5 hours): Severe FPS degradation occurred, dropping below 10 FPS
-> Enabled C-States (from auto to enabled)Flight 2 (3.5 hours): No FPS degradation but experienced heavy MainThread and Manipulator spikes toward the end, causing severe lag spikes
-> Changed preventLabelOverlap to falseFlight 3 (3.5 hours): No FPS degradation and no unusual MainThread or Manipulator spikes
Flight 4 (1.5 hours): No issues either
I don't want to jinx it but I'm optimistic that these two settings fixed it for me
-
I am also having these performance issues after a longer flight. After reading the suggestions above I thought I'd just turn off the MFD by turning off the copilot avionics. And yes, the FPS goes from terrible back to pretty much perfect at an instant. Turning it back on still keeps good framerate but they just get progressively bad again.
Just thought I report this as this might help narrowing down the search for the cause, though this seems to confirm that you were already looking in the right direction. Hope this helps!
Edit:
A video to demonstrate:
https://www.youtube.com/watch?v=ibnetmmZLx4The recording is limited to 30fps so the difference is not as big as I see on my secreen, but the difference is still noticeable in the video.
-
I am also having these performance issues after a longer flight. After reading the suggestions above I thought I'd just turn off the MFD by turning off the copilot avionics. And yes, the FPS goes from terrible back to pretty much perfect at an instant. Turning it back on still keeps good framerate but they just get progressively bad again.
Just thought I report this as this might help narrowing down the search for the cause, though this seems to confirm that you were already looking in the right direction. Hope this helps!
Edit:
A video to demonstrate:
https://www.youtube.com/watch?v=ibnetmmZLx4The recording is limited to 30fps so the difference is not as big as I see on my secreen, but the difference is still noticeable in the video.
@Zeemon That definitely does sound promising! Without changing any other settings on your computer, could you please try the first, very easy suggestion in my post here? With such definitive results on your system, I might hopefully be able to include this in my next update this week. Thank you so much for commenting!
