So SAC and I were discussing performance and pretty code via DM and the question came up wheter a lot of maps or a single map filled with structs with lots of attributes would be more performant.

Since we didnt know, I wanted to search some knowledge here.

Will a script be more performant if I use e.g. 100 std::maps to store and retrieve information for 1000 keys or one single map filled with 1000 keys, each value being a struct with 100 attributes?

Edited January 3, 20223 yr by HughJanus

Let me preface my reply by saying that it is irrelevant in comparison to the performance cost of other operations, such as making native calls.

 

However, the cost of performing map lookup in comparison with a dereferencing a pointer is a lot greater, so a single map with lots of values would be faster. But again, we are talking infinitesimal differences. The degree of organization a single map provides is also worth a lot.