Now, the memory is using BtreeMap is get the base, the algorithm is O(log n) and only need to traverse down 2-3 level for 1000 nodes.

But running the intel vtune profiler for engima.bin show that most of the time is doing btree range search which can take a lot of time and slow down up to 2 seconds.

So maybe we can implement a better approach like bochs, which is a fast hashmap for memory region?

Or maybe fast stack region for faster access to stack.