I'm enjoying using the library! However I can't see any speedup cf regular python, so potentially I'm missing something here. Actually, using the matrix_multiply function from the user manual, I see ...

My computer only has 2 cores with 4 threads each: Square matrix of: 256x256 ~12.1 seconds. Increasing that by 1 bit to 512x512 ~105 seconds (~2 minutes). As I'm running this program on my own Linux OS ...