Solved MFDE impacts FPS with U3
-
I noticed a big change with U3 regarding FPS when MFDE is running …about 30-40fps more if I turn off MFDE…whereas there was any change with U2.
FYI: RTT is also running for displaying the MFDs
-
@suhkoi69 I am not sure what you’re expecting from BMS here…
To be honest, this part is the best optimized yet but we can’t really make things better unless we change the RTT architecture (which isn’t something planed in our books for a while)
Maybe you can try to ask @oakdesign if he can help with his portable version…
Works for me!
-
To be honest, I don’t have such dramatic impact…
MFDE needs to be updated and you can bug Lightning to continue or resume the work…I am using a version compiled by @oakdesign and it works fine!
RTT export will cause fps drops but 30 to 40 fps is exaggerated.
Cheers
-
I mentioned here it’s rather a regression when I compared with U2. It is not as smooth as before and this change is visible : 30fps is not exagerated
However without MFDE, I also noticed the FPS are better than before -
@suhkoi69 We don’t maintain MFDE…
If we had to look for all 3rd parties tools, we would never release
The 3rd parties need to follow our changes, not the other way around… -
@suhkoi69
It’s just a question of comfort…it’s still by far playable -
-
-
@suhkoi69 What version of Windows… and which mode (ie. borderless vs fullscreen) and what screen resolution / how many monitors?
The diff from 120 to 87 fps is about 3ms. That’s about the std dev of frame-timing I observe in BMS…
NVidia FrameView tool might be helpful, to collect detailed timing data to compare.
https://www.nvidia.com/en-us/geforce/technologies/frameview/
(works for AMD graphics too, I think)I realize it’s not easy to flip back and forth from U2 to U3 but might be necessary to rule out some change in Windows, DWM or NVidia driver…
-
@airtex2019 said in MFDE impacts FPS with U3:
The diff from 120 to 87 fps is about 3ms. That’s about the std dev of frame-timing I observe in BMS…
NVidia FrameView tool might be helpful, to collect detailed timing data to compare.Hi
I use:
Windows 10 professionnal
borderless
main monitor ( resolution 5120x1440p) ultra wide curve screen + GSync
second usb 14" external monitor for MFD & gauges (resolution 1600x900p)
GEFORCE RTX2700I will do the test with NVidia FrameView tool …
othewise, yesterday I did several tests and I noticed some interesting points:
- with YAME64 -> FPS are better but when I add fuel quantity gauge …fps drops
- with MFDE , whatever the number of outputs, fps is very low as I mentioned
- g_nRTTExportBatchSize 1 or 2 has a big impact now on fps …5 'd be a good compromise
- AntiAliasing in BMS FALCON graphic menu is always active… despite I set “Override any application setting” in NVIDIA control panel …I can see some change if I put 8 or 1 …this parameter impacts FPS too
-
-
@suhkoi69 I’m not on a g-sync system right now so I can’t get a screenshot of the setting… but go to Start / “Graphics Settings” panel in Windows and see if you have an option for “use Variable Refresh Rate for other apps” … sorry can’t remember the exact wording. (And maybe yours will be en francais, anyway.
See if that makes a difference, on vs off, when running apps that overlay the BMS window.
(Just grasping at straws … I think this setting means DWM will wait for other apps to render, before blitting everything to the front-buffer … it’s the kind of thing that could easily add +3ms when an overlay is present. But … no idea or explanation why that timing behavior would change from U2 to U3.)
Also doublecheck Nvidia settings, for g-sync enabled for both fullscreen and windowed.
-
@suhkoi69 I am not sure what you’re expecting from BMS here…
To be honest, this part is the best optimized yet but we can’t really make things better unless we change the RTT architecture (which isn’t something planed in our books for a while)
Maybe you can try to ask @oakdesign if he can help with his portable version…
Works for me!
-