Metro Monday 3

It’s been two weeks since the prior Metro Monday (these are for announcing changes, and are not strictly weekly). Here’s what’s happened in that time:

On the asset side, there are two new buildings. First, “Anderson Antimicrobial” is a 3x3 tile chemicals facility. The archetype for this building is “level 3” of “consumer goods,” found in dense industrial areas. Second, “Monarch’s Concourse” is a 2x2 tile corporate office highrise. The archetype for this building is “level 3” of “office work,” found in dense commercial areas.

On the code side, while there are plenty of things to keep improving (don’t worry, we have a huge list of stuff to get to — eventually), focus has been primarily on simulation performance. So, the simulation engine now includes a profiling framework, and after setting up plenty of labeled “measurements” within the sim engine code, we can meaningfully identify expensive operations with precision. These measurements are disabled by default since they incur a performance impact.

From profiling the operations involved in savefile loading, the time it takes to load save files has been dramatically improved — for large saves, it is down to as little as a few percent of the previous load time. 

Efficiency past startup, during the normal course of the game, is also improved. Due to improvements in a variety of mechanisms, from pathfinding, to economic re-evaluation queueing strategies, to more nuanced path removal when changing the transit prototypes of a tile, the economic engine is able to handle larger, denser cities without slowing down as much. So, you should be able to play in a world for longer before getting bored due to the sim engine spawning buildings too infrequently.  By no means are these optimizations the end of the road, but they already help to make the economy seem noticeably snappier, especially when there is a very complex transit network layout. Further, these are “true” optimizations that do not just throw more CPU threads at a problem, and thereby may be seen to reduce power consumption by the game in some situations — which is something we do care about, on behalf of those who play on laptops. We may improve scaling across CPU cores later on, but our favorite optimizations are the kinds that improve the experience on hardware of any price, i.e. inclusive of lower-spec machines.