Path: csiph.com!x330-a1.tempe.blueboxinc.net!newsfeed.hal-mli.net!feeder1.hal-mli.net!news.glorb.com!news2.glorb.com!news-out.octanews.net!indigo.octanews.net!auth.brown.octanews.com.POSTED!not-for-mail From: Paul Rubin Newsgroups: comp.lang.python Subject: Re: Python CPU References: <01bd055b-631d-45f0-90a7-229da4a9a362@t19g2000prd.googlegroups.com> <8vps7tF9vuU1@mid.individual.net> <4d97f125$0$29992$c3e8da3$5496439d@news.astraweb.com> <4d981eb5$0$10581$742ec2ed@news.sonic.net> <8vtbclF7q4U1@mid.individual.net> Date: Mon, 04 Apr 2011 02:23:46 -0700 Message-ID: <7xoc4mmkt9.fsf@ruckus.brouhaha.com> Organization: Nightsong/Fort GNOX User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux) Cancel-Lock: sha1:41CapXssP4IZ3FqAntoO4vlCe/Q= MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Lines: 8 NNTP-Posting-Date: 04 Apr 2011 04:23:46 CDT X-Complaints-To: abuse@octanews.net Xref: x330-a1.tempe.blueboxinc.net comp.lang.python:2559 Gregory Ewing writes: > What might help more is having bytecodes that operate on > arrays of unboxed types -- numpy acceleration in hardware. That is an interesting idea as an array or functools module patch. Basically a way to map or fold arbitrary functions over arrays, with a few obvious optimizations to avoid refcount churning. It could have helped with a number of things I've done over the years.