ANN: NumExpr 2.6.5

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

ANN: NumExpr 2.6.5

Robert McLeod
==========================
 Announcing Numexpr 2.6.5
==========================

Hi everyone, 

This is primarily an incremental performance improvement release, especially 
with regards to improving import times of downstream packages (e.g. 
`pandas`, `tables`, `sympy`).  Import times have been reduced from ~300 ms 
to ~100 ms through removing a `pkg_resources` import and making the `cpuinfo`
import lazy.

The maximum number of threads is now set at import-time, similar to `numba`, by 
setting an environment variable 'NUMEXPR_MAX_THREADS'.  The runtime number 
of threads can still be reduced by calling `numexpr.set_num_threads(N)`. 

DEPRECATION WARNING: The variable `numexpr.is_cpu_amd_intel` has been set to a 
dummy value of `False`. This variable may be removed in the future.

Project documentation is available at:


Changes from 2.6.4 to 2.6.5
---------------------------

- The maximum thread count can now be set at import-time by setting the 
  environment variable 'NUMEXPR_MAX_THREADS'. The default number of 
  max threads was lowered from 4096 (which was deemed excessive) to 64.
- A number of imports were removed (pkg_resources) or made lazy (cpuinfo) in 
  order to speed load-times for downstream packages (such as `pandas`, `sympy`, 
  and `tables`). Import time has dropped from about 330 ms to 90 ms. Thanks to 
  Jason Sachs for pointing out the source of the slow-down.
- Thanks to Alvaro Lopez Ortega for updates to benchmarks to be compatible with 
  Python 3.
- Travis and AppVeyor now fail if the test module fails or errors.
- Thanks to Mahdi Ben Jelloul for a patch that removed a bug where constants 
  in `where` calls would raise a ValueError.
- Fixed a bug whereby all-constant power operations would lead to infinite 
  recursion.

--

_______________________________________________
NumPy-Discussion mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/numpy-discussion