Open endolith opened 1 year ago
Generally should use float64/int64 or float32/int32 (SIMD) for speed? Not uint8 or whatever. Deal with memory usage by chunking, not by using weird types.
Default on work computer is np.int32 and np.float64
Generally should use float64/int64 or float32/int32 (SIMD) for speed? Not uint8 or whatever. Deal with memory usage by chunking, not by using weird types.
Default on work computer is np.int32 and np.float64