Testing TSW

[_o_OOOO_oo-Kanawha]

Our Czech expert has created some tests: http://www.trainsim.cz/?mod=article&nod1=33, download link: http://www.trainsim.cz/2/tsw---10x4400- ... zu/2341/0/

"Do you want to test what your computer can do? :) I have prepared a timetable and a scenario (again just a free ride over CSX:HH) with a totally long set of 10x diesel 4400 plus 300 loaded cars from SFJ addon. It all takes place on the basic SPG track (I used SFJ cars because they can simulate the proper weight). I made that every car is loaded with 120 tons + the weight of empty car will give you a pretty good colossus. Personally, I got an incredible 10 FPS while testing (although I was running a live program in the background) :) Otherwise, you will need an SFJ addon for proper functionality."

So, 10 locomotives and 300 cars in scene is not uncommon in typical US railroading, but TSW doesn't seem to be able to handle it? Because apart from FPS, all those Simugraph stuff, sounds, emitters and scripts also need proper CPU thread priority and clock tics to function properly.

Best try on your game if you can run that single train and be satisfied with your computer's performance :-)

[buzz456]

Ahh I don't think ten locomotives and three hundred cars is all that common in American railroads.

[bnsfsubdivision]

Ahh I don't think ten locomotives and three hundred cars is all that common in American railroads.

Not on a single train, in a full scenario? Even more than that.

[buzz456]

Ahh I don't think ten locomotives and three hundred cars is all that common in American railroads.

Not on a single train, in a full scenario? Even more than that.

I guess I just didn't understand what he meant.

[GreatNortherner]

Hi,

Is there any confirmed source that the AI in TSW uses the same physics and simulations as the player? In many games that have AI (or NPC) interaction, the AI "player" uses simplified simulation models to save computing resources. Without knowing how this works in TSW, this testing scenario can really only tell us how well our computer handles rendering that given (human) player train and the related physics/script/... calculations. We probably shouldn't try to make estimates about AI performance based on that.

My

Cheers
Michael

[_o_OOOO_oo-Kanawha]

AI probably uses simple physics and only the basics of scripting. Haven’t played TSW much of late to remember if sound and particle emitters are scripted to engine load for instance.

There is also a testing scenario available with one train of 90 cars, loaded and appropriately weighted apparently: http://www.trainsim.cz/2/tsw---scenar-90-vagonu/2339/0/
In real operations, such a train probably has a DPU at the rear.

Without a train make up program we cannot test for ourselves.

DTG’s optimizations so far have lead to AI trains being shortened, yards being emptied.
So, what does that tell you about UE4’a power and DTG’s programming skills vs original ambitions of TSW at launch?

There have been a few players who collected all the available cars on NEC:NY and perhaps the British or German routes and found out the frame rate drops considerably when trying to run their longest possible train.

[DigitalRails]

TSW is a console game now.
Sad

[_o_OOOO_oo-Kanawha]

From the Czech guy's latest findings on hacking the Timetable:

"So far, the last piece I did are timetables. These are used in both scenarios and Timetable (Service) mode, which is the 24-hour mode. And right here I pause a little. Thanks to the dissection of timetables, I found out (who does not want to have a bad dream of a perfectly functioning dispatcher so do not read further) that the timetable is completely scripted. Every movement of the train, every light signal or switch is pre-set and switches in advance at a certain time even though nothing goes there. The game has scripted all train positions in the timetable including info on which track will be at what moment. I tried to modify the service on my connected RRO RSN track first - the service is displayed like this - but in more detailed display it is everywhere red and the track is not set up at all, because I did not modify the complete route of the set. By the way, my service starts on an unused siding in the RRO, where you just can't get there. Unfortunately, just because of the path set set ends up in the siding and the main move simply does not get (through the editor it would be much easier, of course, it all shakes itself and sets everything - scripts - for proper functionality). I'll have to play with that a little bit. I naively thought that I would enter A and B and the game itself (that is, the intelligent dispatcher) would create a clear path for me;"

Makes you think Free Roam isn't that easy to implement

In the Comments section, the author writes:

"to Callum: merging route is working fine if you have all other hacks which i have. Otherwise its not usable as second part of the route is empty and of course services on the right part failed too as they dont have their final destination and their right starting points. There is no way right now to use two timetables for that. And last think is ... red signals everywhere. So until i figure out how to do right services from RRO to RSN its not usefull and im not gonna waste time to write long instruction how to merge RRO and RSN if there is no service for that right now... when DTG release their tools it will be possible to create services from RRO to RSN (in that time i will write instructions or release tool which will merge this two routes)."

From: http://www.trainsim.cz/?mod=article&sho ... y=#discuss


If and when the Editor arrives, we might perhaps find out for ourselves. Regarding this Editor: DTG is keeping awfully silent on their own e-media outlets.
Not even their preferred business partners have given any indication of producing DLC for TSW.
The fanbois are the only ones talking in circles round Peddlesden. Any other online forum on TSW is like a ghost town filled with zombies.

Glad there still is TS2020, which despite its many flaws allows you to be creative.