Is there a correlation between map size & density, object variety & complexity and crash risk?

[sbaros]

Early this year I started developing a fictional base map to use as a test bed for various industry recipes and other stuff under consideration, usable on all Tycoon versions. I dimensioned it at the maximum perceived size of 1024x1024 . While adding new cities to the map, I gradually enhanced the object variety & complexity too, notably with the building extensions/foundations, but also with additional motive power, such as the self-propelled railcars and many user-provided ones from the forum.

Near the end of the summer, when I had added numerous cities on the map and also building extensions, I started experiencing crashes at an increasing rate during the buildings' seeding phase. Such crashes never happened though when loading the map directly in editor mode. As a first reaction, I immediately halted further building extension development, and I also moved several building extensions out of the UserExtraContent of the affected installation, a portable one for the record. There was some improvement, albeit a limited one in the frequency of crashes. However, I did notice a significant variation from computer to computer. Strangely enough, the newer the computer, the bigger the problem, despite better processors and RAM sizes.

The Windows version itself doesn't look like an important factor either. Both the best (oldest) and the worst-behaving PC (newest) feature Windows10 ! As far as other computers were concerned, my Windows8 laptop crashes quite badly too, while an older Windows7 desktop crashes very rarely and constitutes by far my favourite.

The main question is whether the situation is caused by any of the suspected factors (map size and density, object variety and complexity) or a combination thereof. Have others experienced comparable situations?

[Gumboots]

It's not something I have ever noticed*, but it does make a certain amount of sense that a maximum sized map in combination with a lot of custom building files might tend to overload the game engine during seeding calculations.

*Although I usually avoid 1024x1024 maps anyway, I find them too awkward to manage on screen during play.

[sbaros]

I guess there is no way to "crop" a GMP map's dimensions or "split" it to partial sub-areas without losing all work done on it, right ?

[Gumboots]

It is probably possible in theory, if you do enough direct code editing inside the .gmp, but it's not something I'd like to try.

However, you could easily split the original heightmap, and then add in the stuff for each region fairly quickly.

[sbaros]

Well, the height-map has hardly been 1% of the total job, not an option for me ...

[sbaros]

I have to report a near-disappearance of the aforementioned crashes during the past month or so (no, no April fool!). Since I modified nothing myself, I guess this is to be attributed to Windows bug-fixes that take place automatically and unnoticed. Maybe earlier Windows updates had caused the instability in the first place, whereas my non-updateable Windows7 had remained unaffected all the time.
All's well that ends well.