[Csantucci]

As announced in this forum, the actual official Testing release supports Monogame.
For the sake of uniformity, I tried to apply the official Monogame-related code to OR NewYear MG. However, at least on my computer, I got lower frame rates (and this is a problem for me, because I have a slow computer).
I show here a comparison between the picture I get with the Testing release X.308
 ORTestingX308.jpg (372.16K)
Number of downloads: 36

and the same with release NewYear 92
 ORNYMG92_32.jpg (365.03K)
Number of downloads: 36

As can be seen, the difference is significant. In this particular case, deselecting the "Run at 32 bit on Win64" would lead to an even bigger difference, but that wouldn't be a fair comparison.
Here I attach the two logfiles, to show that the comparison is done with equivalent settings
 OpenRailsLog308.txt (16.71K)
Number of downloads: 10
 OpenRailsLog92.txt (16.96K)
Number of downloads: 10

The only thing I noticed is that in the official release the frame rate drops if option "Glass on in-game windows" is selected.

So by the moment I stick with the MG interface used in OR NewYear MG (even if switching later could be more cumbersome), and would be happy to know if this is a specific problem of my computer or not, or if I did something wrong in the comparison.

[cjakeman]

Hi Carlo,

Csantucci, on 04 March 2021 - 02:35 AM, said:

For the sake of uniformity, I tried to apply the official Monogame-related code to OR NewYear MG. However, at least on my computer, I got lower frame rates (and this is a problem for me, because I have a slow computer).

I agree - 25 to 19 is a poor outcome.

I've tried to reproduce this using 308 and 92 on a slow PC and I get 30 FPS on both versions. Is there an option I should change before switching from Testing to OR_NewYEar_MG ?

 OpenRailsLog308.txt (13.35K)
Number of downloads: 1

 OpenRailsLog92.txt (13.65K)
Number of downloads: 1

[Csantucci]

To perform a comparison at equivalent conditions, you should set the Multisampling count slider to 2 in the ORNYMG version. The testing version has the Multisampling count fixed to 2, because this is the default of XNA (see James' comment at line 121 of RenderProcess.cs, master branch). Furthermore (but this should have a very small influence) you should set the Simple Physics to true (again in the ORNYMG version).
Moreover you should set the "Glass on in-game windows" option to true in both versions, because it's an option that reduces framerate in the Testing version, and I assume that it's a widely used option.
This way you should see a difference in the FPS.

[perpetualKid]

Csantucci, on 04 March 2021 - 11:54 AM, said:

To perform a comparison at equivalent conditions, you should set the Multisampling count slider to 2 in the ORNYMG version. The testing version has the Multisampling count fixed to 2, because this is the default of XNA (see James' comment at line 121 of RenderProcess.cs, master branch). Furthermore (but this should have a very small influence) you should set the Simple Physics to true (again in the ORNYMG version).
Moreover you should set the "Glass on in-game windows" option to true in both versions, because it's an option that reduces framerate in the Testing version, and I assume that it's a widely used option.
This way you should see a difference in the FPS.

I could confirm this behavior on my machines as well. Running Unstable or Testing version of OR, framerate is 20-40% less than ORNYMG, less difference on the machines with more powerful GPU.
Similar behavior compared to ORTS-MG ("OR Ultimate"), but thats a different story as it's much more aggresively optimized...

[Csantucci]

Hi Steffen,
thanks for confirmation. What do you mean with ORTS-MG? Your Ultimate OR version?

[perpetualKid]

Csantucci, on 04 March 2021 - 12:50 PM, said:

thanks for confirmation. What do you mean with ORTS-MG? Your Ultimate OR version?

updated in my post, but yes.

[cjakeman]

Csantucci, on 04 March 2021 - 11:54 AM, said:

To perform a comparison at equivalent conditions, you should set the Multisampling count slider to 2 in the ORNYMG version. The testing version has the Multisampling count fixed to 2, because this is the default of XNA (see James' comment at line 121 of RenderProcess.cs, master branch). Furthermore (but this should have a very small influence) you should set the Simple Physics to true (again in the ORNYMG version).
Moreover you should set the "Glass on in-game windows" option to true in both versions, because it's an option that reduces framerate in the Testing version, and I assume that it's a widely used option.
This way you should see a difference in the FPS.

Thanks for the detailed instructions, Carlo.

I now get 30 => 27 (for OR_NewYear_MG => Testing), so a drop of 10%.

[pschlik]

Had a try at this myself, using the MSTS demo route I was able to achieve a bit under 220 fps in New Year revision 92 (not updated to 92.1), while in the most recent unstable version I only got around 170 fps, so definitely a substantial difference.

With the glass on windows disabled, I get the same kind of results so I don't think it's connected to that. (By the way, I generally leave this option off as I dislike the effect.)

I noticed that New Year is using more of my GPU, but about the same CPU as the official versions. Something might be bottlenecking the official version more than New Year.

[James Ross]

Csantucci, on 04 March 2021 - 02:35 AM, said:

So by the moment I stick with the MG interface used in OR NewYear MG (even if switching later could be more cumbersome), and would be happy to know if this is a specific problem of my computer or not, or if I did something wrong in the comparison.

Partial investigation (more to follow)

I have done some analysis of the performance of the Stable Version, Testing Version, and NewYearMG version on my two computers: a laptop with integrated graphics, and a desktop with discrete graphics.

• Discrete desktop, default* settings: Stable 172, Testing 370, NewYearMG 298
  
• Discrete desktop, high* settings: Stable 163, Testing 371, NewYearMG 270
  
• Integrated laptop, default* settings: Stable 118, Testing 121, NewYearMG 133
  
• Integrated laptop, high* settings: Stable 61, Testing 55, NewYearMG 78

What's interesting here is that NewYearMG is performing better on integrated graphic, but the Testing Version is ahead on discrete graphics.

My guess at this point is that NewYearMG is doing something better on the GPU/graphics side, while the Testing Version is doing something better on the CPU/computation side (it may even be one thing that changes the CPU/GPU balance).

Settings used:

• Default settings are /MultisamplingCount=2
  
• High settings are /MultisamplingCount=2 /ConditionalLoadOfDayOrNightTextures /DistantMountainsViewingDistance=60000 /DynamicShadows /ModelInstancing /PreferDDSTexture /ReduceMemory /SaveCommands=False /SimpleControlPhysics /ViewDispatcher /WindowGlass

I will do further analysis tomorrow into the differences.

[Genma Saotome]

Time and curiosity allowing it might be interesting to compare .dds vs. ace. When I checked that some very long time ago .ace won hands down.