`numpy.exp` is not raising overflow error for large values

Here is an example:

import numpy as np
np.seterr(all='raise')
np.exp(750.0)

this used to raise FloatingPointError: overflow... in older NumPy version, but isn’t raising it as of version 1.14.2

[oleksandr-pavlyk]

@tadeu It will be inefficient to use Intel Math Kernel Library’s VML routines for a single entry due to an overhead cost.

I can confirm that NumPy 1.17 from intel channel does raise the FP exception:

Python 3.6.9 |Intel Corporation| (default, Sep 11 2019, 11:50:16) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
Intel(R) Distribution for Python is brought to you by Intel Corporation.
Please check out: https://software.intel.com/en-us/python-distribution
>>> import numpy as np
>>> np.seterr(all='raise')
{'divide': 'warn', 'over': 'warn', 'under': 'ignore', 'invalid': 'warn'}
>>> np.exp(750.0)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
FloatingPointError: overflow encountered in exp
>>>

However, NumPy in the Intel channel is compiled with Intel C Compiler, and uses Intel’s C compiler run-time, which implement scalar transcendental functions.

NumPy installed from conda-forge channel, presumably also compiled with MSVC raises the error:

(np117_v) C:\Users\opavlyk\devel>python
Python 3.6.7 (default, Jul  2 2019, 02:21:41) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
>>> np.seterr(all='raise')
{'divide': 'warn', 'over': 'warn', 'under': 'ignore', 'invalid': 'warn'}
>>> np.exp(750.0)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
FloatingPointError: overflow encountered in exp

Perhaps someone from Anaconda team needs to look into the problem and comment.

[mattip]

The answer to your question is no, this was a compilation using no linalg accelerator. Additionally, BLAS/Lapack is not involved in np.exp.