Discussion of work on improving the performance of modifications

**youneuoy** · July 05, 2020, 04:57 AM

Here I would like to start a discussion of various things that greatly affect the performance of the game.

To begin with, I suggest talking about scripts - some of them quite strongly affect the performance of the game (waiting time for the next turn, etc.) and are very much in demand in various modifications - we could try replacing them with something else using this project . If anyone knows the "heavy" scripts that are needed for minor changes, you can write about them here.

**franky317** · July 05, 2020, 07:04 AM

surely the heaviest scripts are those where there are many conditions.

example:

Spoiler Alert, click show to read:

this is a part of the city name change script, consider 199 cities for 31 factions.
Which then come in this case are grouped into cultures ....

in the bellumcrucis submod, another quite heavy script is Crusades' Limits Script.

Where through various counters the crusades are identified, setting that only some factions can participate.
All this for not having an unbalanced campaign but it is very heavy.

example:

Spoiler Alert, click show to read:

this is only the script for the crusades against baghad caliphate, zengidi.

I'm already working on this through the label function. if you want I can explain :-)

**Erken** · July 06, 2020, 01:28 AM

Maybe the script for named forts could be replaced with something simpler. Instead of characters spawned and killed each turn, names could be associated in a config file with coordinates, and thus simply displayed IG over designated tiles without any character.

**Lord Baal** · July 08, 2020, 02:27 PM

Now that forts can be made permanent perhaps it would not be necessary anymore and indeed could be named out of a config file.

**Gray Fox** · July 27, 2020, 08:34 PM

Is there anyway to expand on m2tw taking more resource from multi core processors ? I would like to see 20,000 v 20,000 troops / and larger tactical map on a modern alien computer without having lag etc. It would truly make m2tw epic.

**Araval** · July 29, 2020, 12:40 PM

The game checks a monitor_condition in script 20 times per second which makes end turn times long for mods that have this kind of script a lot. Maybe once per second would be sufficient?

**youneuoy** · July 29, 2020, 10:56 PM

Originally Posted by Araval

The game checks a monitor_condition in script 20 times per second which makes end turn times long for mods that have this kind of script a lot. Maybe once per second would be sufficient?

I cannot do exactly that. But if there are some very heavy and at the same time very necessary scripts using monitor_condition, then I can try to completely transfer their functionality to the program.

Gray Fox,
I think it's almost impossible. I use multithreading to write the program, but for the things in the game it would probably be crazy to try to add this.

**Erken** · July 30, 2020, 12:21 PM

The main cause of long scripts is the lack of adequate (independent) conditions. Could the tool do something like giving a value to an event_counter depending on conditions checked out of the game? For example, if a region is local, then an associated counter would have 1 for value, and we would not have to test the settlement owner with each faction:

Code:

if I_EventCounter T_RegionIsLocal_81 = 1

    [...]

end_if

("T" is just for "tool")

Something more complicated could also be done by linking settlement levels with different values for the same counter:

Code:

if I_EventCounter T_SettlementLevel_81 = 5 ; large city

    [...]

end_if

Just an idea.

**youneuoy** · August 14, 2020, 02:17 PM

Originally Posted by Erken

The main cause of long scripts is the lack of adequate (independent) conditions. Could the tool do something like giving a value to an event_counter depending on conditions checked out of the game? For example, if a region is local, then an associated counter would have 1 for value, and we would not have to test the settlement owner with each faction:

Code:

if I_EventCounter T_RegionIsLocal_81 = 1

    [...]

end_if

("T" is just for "tool")

Something more complicated could also be done by linking settlement levels with different values for the same counter:

Code:

if I_EventCounter T_SettlementLevel_81 = 5 ; large city

    [...]

end_if

Just an idea.

I think it's possible to do this, but I'm not sure about expediency.
I need to think carefully about the further direction of the project, perhaps a more correct solution would be to give modders the opportunity to write c ++ code, there are no restrictions and the project already has many different auxiliary functions, rid of low-level code.

For example, a small code that will kill the ruler of the england faction if the capital of the faction is a level 5 castle:

**Erken** · August 15, 2020, 02:04 AM

I know nothing to c++, but your example is very easy to understand, and personally i would prefer learn the rudiments of this langage rather than continuing to suffer the limitations of the campaign_script

.

So if i correctly understand, it's more interresting for you to develop ideas allowing you to make new functions, so that modders can next do what they need with them?

Another question

: The tool perform his actions before the local turn currently, is this a necessity or could it also be done in other circumstances like we do in scripts with monitor_event?

**Lord Baal** · August 15, 2020, 01:24 PM

Having scripting capabilities like that alone would not only revolutionate it all, but even increase performance by making huge and large scripts only a few lines long.

**youneuoy** · August 15, 2020, 01:43 PM

Originally Posted by Erken

I know nothing to c++, but your example is very easy to understand, and personally i would prefer learn the rudiments of this langage rather than continuing to suffer the limitations of the campaign_script

.

So if i correctly understand, it's more interresting for you to develop ideas allowing you to make new functions, so that modders can next do what they need with them?

Another question

: The tool perform his actions before the local turn currently, is this a necessity or could it also be done in other circumstances like we do in scripts with monitor_event?

yes, the main problem is finding all these things. It is difficult to do this when many other things have to be done in parallel. In general, this can be done. Nevertheless, there are some events even now, not exactly the same as usual in scripts, but still. For example, you can write your own code that will be executed when the campaign is loaded or when it is closed. You can also create a separate thread, and make a while loop in it, thus the function of replacing resource models works, for example(this function is about 100 lines long).
Maybe after update 1.19 I'll start working on it. I will make a separate .dll and let modders try to write C ++ code, if they like it, we will expand these capabilities. The only drawback, in my opinion, is that you cannot just write the code into the file, it will also need to be compiled.

I think it might look like this:
there will be a file with code and with empty functions already written, like:
onChangeTurnNum (), afterAutoResolve (), etc. You can add your own code to these functions. With the help of auxiliary functions, you can do anything, for example, there is already a function that allows you to get a list of factions or a list of armies in battle and also perform various operations with them., etc.

Originally Posted by Erken

So if i correctly understand, it's more interresting for you to develop ideas allowing you to make new functions, so that modders can next do what they need with them?

well, in any case, I have to do this because modders usually do not have such skills (reverse game functions and restoring game data structures). So it would save me some time for me.