AI behavior affected by NVIDIA settings?

[Gecky]

Weird observation i was able to replicate many times when experimenting with fast sync and vsync. I was using a mission in "over the hump" Mig29S campaign (the 2nd I think) where there is a flight of Su-25T that will conduct SEAD against SA-11s to test for smoothness of the settings. For some strange reason the Su-25s move and taxi ahead of my flight (causing collisions with my wingmen too) when Im on normal vsync and without triple buffering (triple buffering off). when in fast sync or vsync+triple buffer, they behave as supposed to, taxiing only after my flight. I've replicated it again and again for half a day, and they keep doing it only in normal vsync w/ triple buffer off. I honestly don't understand how this happens

[deadpool]

Weird observation i was able to replicate many times when experimenting with fast sync and vsync. I was using a mission in "over the hump" Mig29S campaign (the 2nd I think) where there is a flight of Su-25T that will conduct SEAD against SA-11s to test for smoothness of the settings. For some strange reason the Su-25s move and taxi ahead of my flight (causing collisions with my wingmen too) when Im on normal vsync and without triple buffering (triple buffering off). when in fast sync or vsync+triple buffer, they behave as supposed to, taxiing only after my flight. I've replicated it again and again for half a day, and they keep doing it only in normal vsync w/ triple buffer off. I honestly don't understand how this happens

 

(I'm not from ED or associated, just a curious person)

Would be interesting if you could generate a trackfile in the triple buffer off mode and then replay it in triple buffer on mode and see if that changes the playback already.

That would narrow it down to a "runtime" or "decisiontime" incident.

[VirusAM]

I think tracks of both behaviors could be useful

[TOViper]

Weird observation i was able to replicate many times when experimenting with fast sync and vsync. I was using a mission in "over the hump" Mig29S campaign (the 2nd I think) where there is a flight of Su-25T that will conduct SEAD against SA-11s to test for smoothness of the settings. For some strange reason the Su-25s move and taxi ahead of my flight (causing collisions with my wingmen too) when Im on normal vsync and without triple buffering (triple buffering off). when in fast sync or vsync+triple buffer, they behave as supposed to, taxiing only after my flight. I've replicated it again and again for half a day, and they keep doing it only in normal vsync w/ triple buffer off. I honestly don't understand how this happens

 

If you are right ... nice observation!

This reminds me a little of the "VR zoom thing" a few months ago, which was solved nicely by ED meanwhile. Back then, the time for reaching the full zoom was depending on the current frame rate, so if you had 90 fps the zoom in was really fast, but when having 45 fps the zoom in was slower.

Also, in the Viggen we had the behaviour that at low frame rates the trim was moving slowly, and at high fps faster, this was also repaired IIRC.

It sounds to me that the calculations the AI is performing "need enough time" for correctness (or they are somehow even left out completely), which a VSYNC obvously gives. Would be interesting to test on my computer as well, but I have no time for ED's jobs :D

[Gecky]

Ive taken note of the suggestions here, Thanks! will produce the tracks (granting this will happen again, will also inform if my conclusion is false). Sorry that I had limited PC hours this week..

Edited June 6, 2020 by Gecky

[Gecky]

couldn't quite make it happen today. I really wonder now what was causing it then (the SU-25s taxiing before my flight of MiGs). Anyway, am I correct to say that Triple Buffering should not even be doing anything to DCS since it's nor OpenGL?