Hello. I'm using Maya to export all the materials from the scene (most of them have textures) and put them in one single file. I've separated the materials from the mesh so I can assign one material to multiple meshed or to change the material of a mesh when I want to while rendering.

I have a class for Materials that keeps pointers to the resource view after I load and Create the textures from the file.

I name the materials so I can assign them to the meshes when I want to. Like Mesh1->materialpointer = Materials("chest_material");

The thing is that I don't know how to store the material information so I can aces it faster when I iterate in the materials to search for that name string ("chest_material") or even a number;

My first idea was to create an Binary Search Tree (with AVL insertion) but I don't like the idea of data being spread all over the ram when I create the nodes so the next idea was to create a simple vector and make a binary search in it to lookup for the material data.

The problem is if I want to destroy a material and a resource view (in case I don't need that material anymore and release the data from ram) or I need to load a new one (ex: new level needs 3 new materials that are not loaded initially for performance issues) I always have to sort the vector and maybe re-size it or reuse its components.

Any suggestion? I really don't want to use any library.

If I assign a new material for one mesh I have to search the material vector (or a list or whatever) for the name. If I have like 500 materials I should do a binary search so I wont iterate all the materials all the time. The question was if I should make a binary search tree to organise material data for fast insertion/lookup but the texture data will be spread all over the ram (because of different allocations when I create a new material item) or to use a vector and make a binary search but then when I need to resize the vector to add another element I will always have to sort it or create it in another location in ram (as in std::vector where if the new size is bigger the whole vector must be moved in another location so the items would be in continuous memory).

I don't consider a list at all. Its a Binary Search Tree or a vector.

Its 500 materials but every time I construct a mesh I must search for the material. So it would be number_of_meshes * number_of_materials. If I have 500 meshes it would be 500*500 = 250.000 times searching for a string.