Optimize Flow Graph
In this section, we'll use the Performance Analyzer to identify and improve the performance of our graph
Performance Analyzer
Find bottlenecks in your dungeon graph designs and optimize them using the Performance Analyzer. It shows how many attempts were made to build the dungeon.
It is normal for a dungeon to make multiple attempts before converging to a solution. However, if your dungeon design doesn't provide good opportunities for it to converge, they would show up here and you can fix them
Run Performance Analyzer
Open up our flow graph editor and reassign the Module database in the Editor Settings, as we've done previously
Please the Play button on the Performance Stats window
We are running 300 test cases, where we generate random dungeons with current graph setup. Each test case will take a certain number of retries before converging to a solution
The number of retries is represented in the X-axis. The number of test cases is represented in Y-axis
We want to reduce the number of retries our graph takes (so it generates them faster). We have quite a few test cases taking more than 10 retries (orange), and some taking more than 30 retries (red)
Optimize Graph
Lift Modules
Looking at the generated graph, one possible reason is the graph doesn't have much opportunity to grow to other floors (and hence utilize the free nodes) because the lift module design is very constrained. It has to enter and exit from the same side.
We'll fix that by adding 3 more lift modules where the top exit is from the other 3 directions
Existing Lift Module:
The existing lift module we created in the previous section enters and exits from the same side
New Lift Modules:
Create 3 new Lift Modules which exit from the other 3 directions like shown below:
This gives the layout graph more opportunities to grow
Register the new Modules
Open up the Module database editor and add the three entries
Rebuild the Module cache and save the module database
Test Performance
Run the performance analyzer again