Author: mattip <matti.pi...@gmail.com>
Branch:
Changeset: r80272:bd3de357fc95
Date: 2015-10-16 16:42 +0300
http://bitbucket.org/pypy/pypy/changeset/bd3de357fc95/
Log: vectorization is disabled by default, also remove slightly
inaccurate connection between optresult-unroll and minor JIT
slowdown
diff --git a/pypy/doc/release-15.11.0.rst b/pypy/doc/release-15.11.0.rst
--- a/pypy/doc/release-15.11.0.rst
+++ b/pypy/doc/release-15.11.0.rst
@@ -5,7 +5,8 @@
We're pleased and proud to unleash PyPy 15.11, a major update of the PyPy
python2.7.10 compatible interpreter with a Just In Time compiler.
We have improved `warmup time and memory overhead used for tracing`_, added
-`vectorization`_ for numpy and general loops where possible on x86 hardware,
+`vectorization`_ for numpy and general loops where possible on x86 hardware
+(disabled by default),
refactored rough edges in rpython, and increased functionality of numpy.
You can download the PyPy 15.11 release here:
@@ -35,22 +36,26 @@
Availability of SIMD hardware is detected at run time, without needing to
precompile various code paths into the executable.
+The first version of the vectorization has been merged in this release, since
+it is so new it is off by default. To enable the vectorization in built-in JIT
+drivers (like numpy ufuncs), add `--jit vec=1`, to enable all implemented
+vectorization add `--jit vec_all=1`
+
Internal Refactoring and Warmup Time Improvement
================================================
Maciej Fijalkowski and Armin Rigo refactored internals of rpython that now
allow
PyPy to more efficiently use `guards`_ in jitted code. They also rewrote
unrolling,
-leading to a warmup time improvement of 20% or so at the cost of a minor
-regression in jitted code speed.
+leading to a warmup time improvement of 20% or so.
Numpy
=====
-Our implementation of numpy continues to improve. ndarray and the numeric
dtypes
+Our implementation of `numpy`_ continues to improve. ndarray and the numeric
dtypes
are very close to feature-complete; record, string and unicode dtypes are
mostly
supported. We have reimplemented numpy linalg, random and fft as cffi-1.0
modules that call out to the same underlying libraries that upstream numpy
uses.
-Please try it out, especially using the new vectorization (via --jit vec=1 on
the
+Please try it out, especially using the new vectorization (via `--jit vec=1`
on the
command line) and let us know what is missing for your code.
CFFI
@@ -64,12 +69,12 @@
.. _`warmup time and memory overhead used for tracing`:
http://morepypy.blogspot.com/2015/10
.. _`vectorization`: http://pypyvecopt.blogspot.co.at/
.. _`guards`: http://rpython.readthedocs.org/en/latest/glossary.html
-
.. _`PyPy`: http://doc.pypy.org
.. _`RPython`: https://rpython.readthedocs.org
.. _`cffi`: https://cffi.readthedocs.org
.. _`modules`:
http://doc.pypy.org/en/latest/project-ideas.html#make-more-python-modules-pypy-friendly
.. _`help`: http://doc.pypy.org/en/latest/project-ideas.html
+.. _`numpy`: https://bitbucket.org/pypy/numpy
What is PyPy?
=============
_______________________________________________
pypy-commit mailing list
pypy-commit@python.org
https://mail.python.org/mailman/listinfo/pypy-commit
