3 ms·
If the number crunching is numpy-heavy, multiple threads absolutely can use real thread-based parallelism. C extensions like numpy release the GIL and so can ru
by doubleunplussed 5y ago
If the number crunching is numpy-heavy, multiple threads absolutely can use real thread-based parallelism. C extensions like numpy release the GIL and so can run in parallel.