I’m not sure if this is of use to anyone but here you go.
So for one project i am working on, i’m in need of the fastest possible way to create access and modify a dynamic 2d grid of actors, because the creation of these are completely on the fly based on user input. so i looked at two separate community implementations of dynamic containers and built some extra functionality on top of it all to calculate how much time in seconds and milliseconds each could be created and filled with random data. A container called VictoryGrid which is basically a ustruct of a tarrays containing a ustruct of tarrays containing astaticmeshactors from . The other from robber9000, a maze uobject class which is a wrapper for a TMap that looks like TMap< FIntVector, AStaticMeshActor >
so i got the time and stored it in a file, make single container type, and fill it with random data. render it on screen, check if all actors are valid and make sure there are the right number, if so, get time and store it in the file, take the starting time from the finishing time to get the time taken, and store it in the file
if not the right number of actors or invalid actors throw test away (none actually failed)
i wrote a program to calculate the data on the 100 tests i did for each 25x25 50x50 and 100x100 build size, so 300 tests for each container type. 600 tests in total and took an average of the time it took to do each build. i did this semi manually btw >.> play -> stop x 600 times lol
I came up with these numbers, on average ramas had this
Mode - Size x Size = average time it took to complete test in seconds.miliseconds
PIE Mode - 25 x 25 = 0.159
PIE Mode - 50 x 50 = 0.405
PIE Mode - 100 x 100 = 3.906
I came up with these numbers, on average robber9000 had this
Mode - Size x Size = average time it took to complete test in seconds.miliseconds
PIE Mode - 25 x 25 = 0.160
PIE Mode - 50 x 50 = 0.430
PIE Mode - 100 x 100 = 4.112
so not that much speed difference, but it looks like the ustruct tarray of ustruct tarray of astaticmeshactors is faster, even though it took a long time it was an interesting test and i got my answer. you can find ramas code on the wiki and you can find robber9000’s code on his git
bye