Significant FPS drop with last build 34.0.2

[vcaptain747 44]

Hi Nico,

The Fenix TCAS is working now with the last build. Thanks! 

However, I noticed a  significant FPS drop that happens as soon as live traffic is injected, so reverted back to 34.0.0 for comparison and it works well. 

It seems that MSFS doesn't like setting the airspeed simvar periodically on high frequency. I noticed the same behavior when I worked on the POC for the TCAS, and setting the airspeed only once (or each 2 sec or so) for each aircraft seems to give better FPS.

But It could be unrelated also, so maybe others users can confirm if thy are experiencing the same issue.

Thanks.

Amine.

P.S: could not locate the original post so I started a new one.

  

 

 

 

[somiller 694]

What I found interesting was the penalty for 30 live aircraft was a 20 fps drop even on 34.0.0. Before live aircraft injection there was no fps loss on either version. Also interesting is that with live aircraft injected main thread and manipulators were getting hammered, especially so on 34.0.2!

On the second load using 34.0.2 I noticed with 27 live aircraft fps was around 40 - 44, but when the 28th and 29th live aircraft were injected fps tanked.

[SierraDelta 1,337]

I flew EGLL-EGPH earlier today with 34.0.2 and didn’t notice any severe fps drops, maybe a bit of stuttering as I was panning around in outside view at EGLL.

[kiek 1,213]

3 hours ago, vcaptain747 said:

It seems that MSFS doesn't like setting the airspeed simvar periodically on high frequency. 

No I don't think MSFS has problems with that. I have add the TAS and Vertical speed to the block of data (Lat, Lon, Pitch, Bank, Heading and Alt)  that is pushed at each Sim Frame for each aircraft. Vertical speed is needed for the arrows. First I tried to set VS every 5 seconds, just like the TCAS, but then the value does not hold (strangely enough). I have to update the VS at the same frequency as Lat/Lon etc. TAS I added so I do not need a separate process for that. The block of data went from 32 to 40 bytes, not a big deal for todays processors. 😉 

This happens now for each user aircraft in MSFS, so for the PMDG B738 too, the only difference being that, for those aircraft the values for TAS and VS are set to 0, because they do not need it.

I see no performance degradation at all in the PMDG B738, so this process does not give extra significant overhead.   You could check that too....

 

In the Fenix A320 I don' t see performance degradation viewing outside the cockpit. However, when entering the cockpit the fps drops (over here at Asobo EHAM) from 44 to 32 fps. 

 

EDIT: I watched your video, wowww! that must be something else, back to 15?.

Edited January 24, 2023 by kiek

[kiek 1,213]

15 minutes ago, SierraDelta said:

I flew EGLL-EGPH earlier today with 34.0.2 and didn’t notice any severe fps drops, maybe a bit of stuttering as I was panning around in outside view at EGLL.

What aircraft? And did you download 34.0.2 after I had made some changes? So are you really using the latest build?

Edited January 24, 2023 by kiek

[vcaptain747 44]

Quote

In the Fenix A320 I don' t see performance degradation viewing outside the cockpit.

I was using the very basic Grumman Goose aircraft as you can see in the PSXT UI in the video , and the Fenix software was closed anyway .

 

Quote

First I tried to set VS every 5 seconds, just like the TCAS, but then the value does not hold (strangely enough)

That's because as soon as you set up the airspeed, the VS simvar comes alive and MSFS starts calculating it (erratically) based on the TAS value.. VS should be the difference between the previous altitude and current altitude divided by the number of seconds between the two readings.. but for some strange reasons, it takes the TAS into consideration too.

Will continue testing and report back.

Thank You!

Amine.

[kiek 1,213]

12 minutes ago, vcaptain747 said:

That's because as soon as you set up the airspeed, the VS simvar comes alive and MSFS starts calculating it (erratically) based on the TAS value.. 

That should be the case for AI aircraft, but PSXT's aircraft are just aircraft objects. The MSFS AI engine should not do anything with it, I ask MSFS to leave them alone with the 

KEY_FREEZE_LATITUDE_LONGITUDE_SET, KEY_FREEZE_ALTITUDE_SET and  KEY_FREEZE_ATTITUDE_SET events..

[vcaptain747 44]

37 minutes ago, kiek said:

EDIT: I watched your video, wowww! that must be something else, back to 15?  Did you update FSLTL, adding new models/liveries?  I don't trust these guys. Try AIG only, they are much lighter on frames.

in the video , I first started 34.0.2 and frames went from 60 to 15 (went as low as 5 when i was not recording) .. but then you see I started 34.0.0 and frames were steady at 40 . Started 34.0.02 again and went back to 15 

[kiek 1,213]

1 minute ago, vcaptain747 said:

in the video , I first started 34.0.2 and frames went from 60 to 15 (went as low as 5 when i was not recording) .. but then you see I started 34.0.0 and frames were steady at 40 . Started 34.0.02 again and went back to 15 

That does not happen here.

[vcaptain747 44]

What I meant is that i'm using both PSXT versions in the video in the same session , so it can't be FSLTL  models/liveries.

[kiek 1,213]

14 minutes ago, vcaptain747 said:

What I meant is that i'm using both PSXT versions in the video in the same session , so it can't be FSLTL  models/liveries.

Well I have tested over here as well with PMDG and Fenix, but I see no fps problems, not even in the Fenix cockpit with TCAS on and no PFD pop up window.

I will upload my current 34.0.2. Maybe that makes a difference. Give it a try, in a few minutes.

[kiek 1,213]

Available now

[vcaptain747 44]

Thanks ! will give it a try now.

[Silicus 850]

I can confirm that I have NO fps drop with the 34.02 version and FSTL.

Thanks so much for adding TCAS!!!

Edited January 24, 2023 by Silicus

[vcaptain747 44]

I'm getting very mixed results with the version you shared. Can you confirm  that the TAS and VS are only set if the Fenix is being used? 

EDIT: I confirm that the version I downloaded yesterday few minutes after release and the one you shared today are different in terms of file size and also behavior. Maybe there was a change in the code after release?

One key difference is that the version I have sets the TAS and VS no matter what the user aircraft is, which seems to cause the FPS drop. In the other hand, the one you shared sets the TAS and VS only if Fenix is being used, otherwise you would see TAS 0 and VS flickring between random values and 0 . I used SimvarWatcher from the SDK to confirm the observation.

Now if I use any aircraft except the Fenix, then things are smooth, but if  I switch to Fenix, then the FPS drops  are still happening but randomly as opposed to the initial version where it would happens the moment live traffic appears.

Will continue testing tonight and see if there is any pattern..

Thanks! 

Edited January 24, 2023 by vcaptain747