Alex Gaynor avatar Alex Gaynor committed 46deb3f

First pass grammar/english changes.

Comments (0)

Files changed (1)

blog/draft/numpy_roadmap.rst

 Numpy in PyPy - status and roadmap
 ==================================
 
-Hello
+Hello.
 
-Among things people really want to have in PyPy is numpy integration.
-This post tries to describe where we are, what we plan (or what we don't plan)
-and how you can help.
+
+NumPy integration is one of the single most requested features for PyPy. This
+post tries to describe where we are, what we plan (or what we don't plan), and
+how you can help.
 
 **The short version for impatient: there are experiments being done, which are
-already faster and better than numpy, and there
-is a path forward, but there is a definite lack of dedicated people or money
-to tackle that.**
+already faster and better than numpy, and there is a path forward, but there is
+a definite lack of dedicated people or money to tackle that.**
 
 The longer version
 ------------------
 
-The numpy effort in PyPy was my project for a while on an on-and-off (mostly
-off) basis over the past two years. There were `some experiments`_ then
-mostly nothing and then some more experiments that are documented below.
+The NumPy effort in PyPy has, for the past two years, been my on-and-off-again
+project. There were `some experiments`_ then mostly nothing and then some more
+experiments that are documented below.
 
-The general idea that seems to be worthwhile pursuing would be to implement
-numpy in RPython (an implementation language of PyPy) and then leverage
-JIT to achieve extra speedups. The really cool thing about this part is that
-overall JIT improvements will benefit numpy performance out of the box, without
-the extra tweaking. As of now there is branch called `numpy-exp`_ which contains
-a translatable version of a very minimal version of numpy in the module
-called micronumpy. `Example benchmarks`_ show the following:
+The general idea that seems to be worth pursuing would be to implement NumPy in
+RPython (the implementation language of PyPy) and then leverage JIT to achieve
+extra speedups. The really cool thing about this part is that overall JIT
+improvements will benefit NumPy performance out of the box, without extra
+tweaking. As of now there is branch called `numpy-exp`_ which contains a
+translatable version of a very minimal version of numpy in the module called
+``micronumpy``. `Example benchmarks`_ show the following:
 
 +--------------------------------+---------------+-------------+
 |                                | add           | iterate     |
 | PyPy numpy-exp @ 3a9d77b789e1  | 0.120s (2.2x) | 0.087 (48x) |
 +--------------------------------+---------------+-------------+
 
-As you can see, the moment floats cross the numpy-python boundary, PyPy's
-JIT goes blazingly fast, but even running array addition is faster by
-a fair degree (although numexpr is still faster, we're working on it).
+As you can see, the moment floats cross the numpy-python boundary, PyPy's JIT
+goes blazingly fast, but even running array addition is faster by a fair degree
+(although numexpr is still faster, we're working on it).
 
-The exact way how array addition is implemented is worth another blog post,
-but in short it lazily evaluates the expression forcing it at the end
-and avoiding intermediate results. This way scales much better than numexpr
-and can lead to speeding up all the operations that you can perform on matrices.
+The exact way how array addition is implemented is worth another blog post, but
+in short it lazily evaluates the expression forcing it at the end and avoiding
+intermediate results. This way scales much better than numexpr and can lead to
+speeding up all the operations that you can perform on matrices.
 
-The next obvious step would be to extend the JIT to use SSE operations on
-intel x86 architecture which should speed it up by about additional 2x.
+The next obvious step would be to extend the JIT to use SSE operations on x86
+CPUs, which should speed it up by about additional 2x.
 
-Overall it seems pretty obvious that reimplementing numpy in PyPy (in RPython)
+Overall it seems pretty obvious that reimplementing NumPy in PyPy (in RPython)
 can bring most of the useful compatibility within a month-two-three of work.
 It also seems that the result will be faster for most cases and the same speed
 as original numpy for other cases. The only problem is finding the dedicated
-person willing to spend quite some time on that and personally I would be
-willing to both mentor such a person and encourage him or her :)
+persons willing to spend quite some time on this and however, I am willing to
+both mentor such a person and encourage him or her.
 
-Another option would be to sponsor numpy development. In case you're interested,
-please get in touch with us or leave your email in comments.
+Another option would be to sponsor NumPy development. In case you're
+interested, please get in touch with us or leave your email in comments.
 
 Cheers,
 fijal
Tip: Filter by directory path e.g. /media app.js to search for public/media/app.js.
Tip: Use camelCasing e.g. ProjME to search for ProjectModifiedEvent.java.
Tip: Filter by extension type e.g. /repo .js to search for all .js files in the /repo directory.
Tip: Separate your search with spaces e.g. /ssh pom.xml to search for src/ssh/pom.xml.
Tip: Use ↑ and ↓ arrow keys to navigate and return to view the file.
Tip: You can also navigate files with Ctrl+j (next) and Ctrl+k (previous) and view the file with Ctrl+o.
Tip: You can also navigate files with Alt+j (next) and Alt+k (previous) and view the file with Alt+o.