[WCAM-1]

Hello everyone,

I have updated to OR MG Version 51. I’m still facing the issue of camera stuttering in 3D Cab. Other camera views work just fine.
Is anybody else facing a similar issue?

[lineman]

Thank you Carlos!!

[Brace_2011]

Csantucci, on 30 January 2020 - 08:57 AM, said:

So you are saying that, also being in explore mode and doing nothing, after some time the FPS get reduced? And how much approximately is this time? And what is the approximate FPS drop?

Hi Carlo

Thanks for your reply. Yes. that is exactly what I am saying - but I have done some further testing and show the results below.

I have run the tests again using ORNYMG v.49.2 (64 bit), Unstable v.U2020.01.17-0936 and Testing version X1.3.1-107 and for completeness I show the results below. My methodology in each case was as follows: turn off logging and dynamic weather etc.; load the game and press [F5] and wait for the FPS to settle; press [ESC] to start the game, but remain in cab view and touch no controls; record FPS after 1 and 2 minutes; after 2 minutes scroll through the F5 HUD to the DEBUG screen and note Updater Process % and FPS (FPS1 in the table below); scroll out of DEBUG screen and note FPS agin (FPS2). This method should ensure that the results are unaffected by AI, loading new tiles, weather or anything else. The only thing 'moving' should, in theory, be the road traffic.

My initial tests were with the Dorset Coast route (notably heavy on resource use) on the path I used in December's tests, and I got the same issues as reported then.

I then completed a second set of tests on the Great Eastern route (a commercial add-on that is pretty light on resources). The results show little difference between ORNYMG and Unstable, but I was a bit surprised by the Testing version results.

So I reverted to Dorset Coast, with the same consist as before but a different path (crucially, a different start point). The road traffic should be the same. Like test 2 the results show little difference between ORNYMG and Unstable.

Comparison between the results for test 1 and test 3 seems to suggest that there is something on the loaded tiles in test 1 that is 'sucking' the FPS out of the game in all the tested versions, but especially ORNYMG - perhaps a dynamic scenery item (a working windmill or something - I can't see anything likely but it doesn't mean it is not there).

To conclude, whatever it is it appears to be a problem that affects ORNYMG more than the others. I am not sure that you will be able to solve the problem without specifically identifying the cause of the issue. Also, I wonder if the respective Updater Process data indicates an area that you might want to look at - or are these results entirely to be expected?

I hope these results help.

Thanks for your patience, and please keep up the tremendous work you are doing to make OR Monogame the game of choice. It is much appreciated.

Regards

Martin

PS. just a final thought Can anyone else with Dorset Coast v6 replicate these results, or is it something in my system or settngs

Attached File(s)

•  FPS tests - 30 Jan 20.jpg (78.21K)
  Number of downloads: 5

[rickloader]

Hi Carlo, thanks very much for your work on OR mg.
With Revision 51 in ctrl 9 dispatch viewer, the trains no longer follow the tracks closely, but "snake" along. Sometimes they are not even on the tracks at all. I think this is a recent problem. The December revision was ok for me
w10 64 Nividia
If I check "run at 32 bit on win64" the trains display correctly
Best wishes, Rick

[Csantucci]

I have uploaded Rev.51.1, that solves the problem of the non rotating wheels.
Rick, your problem is probably a yet unresolved problem of the 64 bit porting. Some problems in the dispatcher window have already been solved, but I didn't notice this one, so I didn't work on it. When I have time I'll have a look to it.

[Diter258r]

Everything ok here

[Csantucci]

Martin,
by chance I have the Dorset Coast route, and so I repeated your tests with your path. My Notebook dates back to 2011, so the FPS are quite different from your ones...
The results are partly equal, and partly different from yours:

OR NewYear MG 51.1 at game start 28 FPS, after 2 minutes 27 FPS
official OR unstable 2020.01.31 at game start 28 FPS, after 2 minutes 27 FPS
official OR Testing 1.3.1-120 at game start 44 FPS, after 2 minutes 37 FPS
The fact that here the OR Testing is faster than the MG versions is due to my poor and old GPU.
Interesting is however that OR testing version has a significant frame drop. So I decided to concentrate on it, and on the possible culprit: the spawned cars.
if you look with TrackViewer at the place where the path starts, you will notice that the 8 surrounding tiles are quite rich at car spawners. So I added to the Debug info of the testing version a line showing how many spawned cars there are.
I have taken 4 pictures, that I attach below: at minute 0, at minute 1, at minute 2 and at minute 3.
At minute 0 FPS are 39, cars are 148;
at minute 1 FPS are 34, cars are 706;
at minute 2 FPS are 33, cars are 932;
at minute 3 FPS are 31, cars are 1002;
cars would then stabilize around 1040.
So I see a direct relationship between increase of spawned cars and decrease of FPS. This occurs in the testing version, so MG is not involved.
The number of cars stabilizes, so there is not a problem of garbage collection. At the moment I stop the analysis here.

Minute 0
 Minute0.jpg (429.49K)
Number of downloads: 2

Minute 1
 Minute1.jpg (432.03K)
Number of downloads: 1

Minute 2
 Minute2.jpg (432.9K)
Number of downloads: 3

Minute 3
 Minute3.jpg (434.64K)
Number of downloads: 2

[Brace_2011]

Hi Carlo

Thanks very much for your patience and attention. I perhaps wasn't a million miles away when I said "The only thing 'moving' should, in theory, be the road traffic". Only, stupidly on my part I thought/hoped this potential source would be eliminated by loading a different path - I thought it might be a rogue shape rather than the volume of road traffic.

Who would have thought that road traffic could have such an impact on FPS - possibly a lesson here for route builders.

Keep up the great work - it is much appreciated.

Regards

Martin

[lineman]

Thanks again Carlos, wheels are working perfectly again!! Thank you for your quick reply.

[ErickC]

Whoa - rolling stock is all going crazy on the rails with this release. Trucks keep jumping all over the place on straight track.

EDIT: I suspect this is actually a track problem and that previous iterations of OR somehow damped the oscillations out, because I always remember vibrations being induced over these particular sections of track. Anyway, I took some screen caps:

 00.JPG (411.75K)
Number of downloads: 2

 01.JPG (413.7K)
Number of downloads: 3

 02.JPG (350.74K)
Number of downloads: 3

The route is the ScaleRail version of Whitefish4_Hawk. I haven't modified it (that black magic is well beyond me!).