Description
Performance is an important part, and we've been working hard to improve the performance of NumSharp, from ArrayStorage to TypedArrayStorage. We are not satisfied with this, we are always looking for a way to make memory alloca
699F
tion more efficient, we propose to use jemalloc.NET to help NumSharp to manage and allocate memory, the new engine will be named UnmanagedByteStorage.
Using unmanaged memory for storage is advantageous for interworking with other Tensor libraries, ideally for zero copy.
Welcome to comments.