OpenRails v360 vs. MSTS performance comparison

[conrail1]

Here is a performance comparison between MSTS and OpenRails v360 with the Feather River Route.

The “Grain Empties” (which uses a 88 car/3 loco/1 caboose consist) and the “South Oroville Switching” activity were used for the comparison. 1600X1200 resolution was used throughout the comparison since it is the highest resolution that MSTS can use.

The system (full specs below) used in the comparison was reformatted with a fresh Windows XP/SP2 64-bit Edition OS, with all of the latest Windows updates and the ATI Catalyst 10.9 drivers.

When the Catalyst 10.10's are released I'll most likely run a similar comparison using Windows 7 64-bit.

ASUS Rampage II Extreme (1802 BIOS)
Intel Core i7-975 Extreme Edition w/Thermalright Ultra-120 eXtreme 1366 RT
Mushkin Redline Ascent 6GB DDR3 1600 (PC3 12800) (6-7-6-18-1N)
Asus/ATI 5870 (Catalyst 10.9)
Creative Sound Blaster X-Fi Titanium Fatal1ty
WD VelociRaptor 150GB Windows XP/SP2 64-bit Edition
WD VelociRaptor 300GB Games/Programs
SILVERSTONE 1500Watt SST
SILVERSTONE TJ09-B



MSTS, 8x Super-Sampling AA, 1600x1200 resolution, all options enabled including dynamic shadows (37 fps) -






MSTS, 8x Super-Sampling AA, 1600x1200 resolution, distant mountains disabled (38 fps) -







OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution (73 fps) -

As I've mentioned before the Super-Sampling AA with ATI and it's 5000 series cards does not perform as well as Nvidia's with the 400 GTX series, but it does an excellent job of cleaning up those transparency textures. It produces some of the sharpest, shimmer free image quality I've seen.





MSTS, 8x Super-Sampling AA, 1600x1200 resolution, all options enabled including dynamic shadows (25 fps) -





MSTS, 8x Super-Sampling AA, 1600x1200 resolution, distant mountains disabled (26 fps) -






OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution (68 fps) -






MSTS, 8x Super-Sampling AA, 1600x1200 resolution, all options enabled including dynamic shadows (25 fps) -





MSTS, 8x Super-Sampling AA, 1600x1200 resolution, distant mountains disabled (25 fps) -







OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution (65 fps) -

[conrail1]

MSTS, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (41 fps) -






OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (92 fps) -





MSTS, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (31 fps) -





OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (74 fps)





MSTS, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (35 fps) -





OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution, dynamic shadows disabled (74 fps)






Here is an example of how Anti-Aliasing and the type of Anti-Aliasing can effect performance.


MSTS, 8x Super-Sampling AA, 1600x1200 resolution, all options enabled (46 fps)







OpenRails v360, 8x Super-Sampling AA, 1600x1200 resolution, all options enabled (75 fps)







MSTS, 8x Multi-Sampling AA, 1600x1200 resolution, all options enabled (46 fps)

As we've known for years MSTS is so CPU dependent that resolution and the amount of AA/AF have little no impact at all on performance -





OpenRails v360, 8x Multi-Sampling AA, 1600x1200 resolution, all options enabled (124 fps)

With Microsoft's XNA Framework the GPU is doing more work compared to what we've seen with MSTS's CPU bound game engine. In this case reducing the AA settings contribute to a major increase in performance.

[sna25]

TEST 1

ORTS Krasnik station, FPS: 71



MSTS Krasnik station, FPS: 28



TEST 2

ORTS Krasnik station, FPS: 74



MSTS Krasnik station, FPS: 21



TEST 3

In the woods, ORTS FPS: 104



In the woods, MSTS FPS: 54



Configuration of ORTS:



Configuration of MSTS:

1280x1024, Everything ON insted of Dynamic Shadows (some polish 3D models crash with DS so I've disabled it.) I also find that in ORTS, dynamic

shadows can be ON and game not crash but for testing I also disable it in ORTS to compare FPS of 2 games with the same configuration.

Computer:

Athlon II X3 3000 MHz
2 GB DDR3
GeForce GT240 512 MB DDR5
Windows Vista Ultimate Box 64 bit.

[GoGoran]

Interesting analysis, thanks for your research !

I must try this with my machine, it's a little older and much less powerful.

At any rate, your test seems to prove that OR performs better on contemporary systems than MSTS does.

But on the other hand, OR is still lacking lots of features like Distant Mountains and Dynamic Shadows, which have yet to be implemented.

[OTTODAD]

Can anybody tell the visual difference between lets say 30 and 70 FpS in MSTS / ORTS, 30 being the frame rates as used by US NTSC TV ?

On my AMD Phenom-2 and ATI HD4870 WIN-XP computer frame rates as low as 20 show smooth action of everything running in MSTS, occasional stuttering when more data is being loaded into RAM unavoidable even with the WD Raptors used by it.

O t t o

[twpol]

But on the other hand, OR is still lacking lots of features like Distant Mountains and Dynamic Shadows, which have yet to be implemented.

Dynamic Shadows are implemented in Open Rails, but there is certainly still plenty for us to do!

[GoGoran]

Dynamic Shadows are implemented in Open Rails, but there is certainly still plenty for us to do!

I was judging from the posted screenshots, which do not show any such shadows. Maybe they were not enabled.

But kudos to you and the OR team - your project is the most important for the MSTS community. I crave the day the first fully-featured and playable OR version will be released. I can't imagine why MS dropped the MSTS ... or why no other company ever tried to create a simulator that is compatible with MSTS content.

[GoGoran]

Can anybody tell the visual difference between lets say 30 and 70 FpS in MSTS / ORTS, 30 being the frame rates as used by US NTSC TV ?On my AMD Phenom-2 and ATI HD4870 WIN-XP computer frame rates as low as 20 show smooth action of everything running in MSTS, occasional stuttering when more data is being loaded into RAM unavoidable even with the WD Raptors used by it.
O t t o

Normally 20-25fps will be fine for a good visual impression, with smooth movement. But when you're running on a track with 20fps average, you're bound to fall way below that rate as soon as heavy traffic or highly detailed surroundings with lots of objects occur (city, train station, yard).

For the sake of saving processing time, OpenRails should cut off the fps at some point, or include that as a scalable option, like ( ) fps unlimited, ( ) fps limit 60, ( ) fps limit 30. That would help to save energy and kee the system cool and quiet as well.

[lindsayts]

Normally 20-25fps will be fine for a good visual impression, with smooth movement. But when you're running on a track with 20fps average, you're bound to fall way below that rate as soon as heavy traffic or highly detailed surroundings with lots of objects occur (city, train station, yard).

For the sake of saving processing time, OpenRails should cut off the fps at some point, or include that as a scalable option, like ( ) fps unlimited, ( ) fps limit 60, ( ) fps limit 30. That would help to save energy and kee the system cool and quiet as well.

I believe I read either on here or UKtrainsim that the OR developers thought that anything over 60 FPS was a waste of processing time as the monitors can only display at a rate somewhere around there and that they would in the end limit the rate to 60 FPS or something like that. But at this stage as it was a help during the development stage they would leave the frame rate display statistic calculations as they currently are.