This is not the most efficient allocator, but it is very simple to implement. The picture below illustrates the process Sequential allocations schematic Every time you want to allocate memory you iterate over each block and check the distance from this block end to the beginning of the next block, if it is large enough you create a new allocation and insert it between those blocks. The way this algorithm works is you have a sorted list of allocated memory blocks, where each entry has two members: offset from a beginning and its size. Sequential fit algorithm was used to manage memory in all those buffers. I didnвЂ™t distinct memory systems, just a few chunks of preallocated GPU memory, that were used for constant buffer allocations and for geometry streaming. Basically, I had a singleton class Renderer which contained 95% of the logic, also to keep it simple IвЂ™ve decided to make it single-threaded for now, so no benefits from using modern API like DirectX12 which are meant to be used with multi-threading. The first iteration of my renderer was as monolithic as it gets.

Re.CinematicSetPalette = DX12_SetPalette

Re.DrawStretchRaw = DX12_Draw_StretchRaw Re.DrawFadeScreen = DX12_Draw_FadeScreen Re.DrawStretchPic = DX12_Draw_StretchPic Re.DrawGetPicSize = DX12_Draw_GetPicSize Re.BeginRegistration = DX12_BeginRegistration

Prototype 2 texture mod