Many of the ufuncs like exp and log are hightly optimized kernels which use SIMD intrinsics where possible. The code can be found in various files in tthe numpy/core/src/umath directory, for instance here is the beginning of the SIMD exp function https://github.com/numpy/numpy/blob/f521ee5f9a9868f38f7ca46d3c6a01aff8bea382/numpy/core/src/umath/loops_exponent_log.dispatch.c.src#L413 Matti
On Fri, Dec 10, 2021 at 3:53 PM <[email protected]> wrote: > > Hello, > > Can someone tell me (or point me to the relevant source code) what algorithms > NumPy uses to compute transcendentals, such as exp, log, etc? > > Thanks, > Jeremiah > _______________________________________________ > NumPy-Discussion mailing list -- [email protected] > To unsubscribe send an email to [email protected] > https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ > Member address: [email protected] _______________________________________________ NumPy-Discussion mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ Member address: [email protected]
