View source for Psyco

{{Short description|Just-in-time compiler for Python}}
{{Distinguish|Psycho (disambiguation)}}
{{Infobox software
| name = Psyco
| logo =
| screenshot =
| caption =
| developer = Armin Rigo, Christian Tismer
| discontinued = yes
| latest_release_version = 1.6
| latest_release_date = {{Start date and age|2007|12|16}}
| programming_language = [[C (programming language)|C]], [[Python (programming language)|Python]]
| platform = [[32-bit]] [[x86]] only
| operating_system = [[Cross-platform]]
| genre = [[Just-in-time compilation|Just-in-time compiler]]
| license = [[MIT License]]
| website = {{URL|psyco.sourceforge.net}}
}}
'''Psyco''' is an unmaintained [[run-time algorithm specialisation|specializing]] [[just-in-time compiler]] for pre-2.7 [[Python (programming language)|Python]] originally developed by [[Armin Rigo]] and further maintained and developed by [[Christian Tismer]]. Development ceased in December, 2011.<ref>{{cite web |title=psyco / Commits |url=https://bitbucket.org/arigo/psyco/commits/all |website=Bitbucket |publisher=Armin Rigo}}</ref>
Psyco ran on [[BSD]]-derived operating systems, [[Linux]], [[Mac OS X]] and [[Microsoft Windows]] using [[32-bit]] Intel-compatible processors. Psyco was written in [[C (programming language)|C]] and generated only 32-bit [[x86]]-based code.
Although Tismer announced on 17 July 2009 that work was being done on a second version of Psyco,<ref>{{cite web |last1=Tismer |first1=Christian |title=[pypy-dev] ANN: psyco V2 |url=https://mail.python.org/pipermail/pypy-dev/2009-July/005288.html |website=pypy-dev mailing list |date=17 July 2009}}</ref> a further announcement declared the project "unmaintained and dead" on 12 March 2012 and pointed visitors to [[PyPy]] instead.<ref name="homepage">{{cite web|url=http://psyco.sourceforge.net/|title=Psyco Homepage}}</ref> Unlike Psyco, PyPy incorporates an interpreter and a compiler that can generate C, improving its cross-platform compatibility over Psyco.
==Speed enhancement==
{{Update|section|inaccurate=yes|date=September 2018}}
Psyco can noticeably speed up CPU-bound applications. The actual performance depends greatly on the application and varies from a slight slowdown to a 100x speedup.<ref>{{cite web |url=http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=psyco |title=Python Psyco benchmarks |accessdate=2008-04-24 |url-status=dead |archiveurl=https://web.archive.org/web/20080606020118/http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=psyco |archivedate=2008-06-06 }}</ref><ref>{{cite web |url=http://psyco.sourceforge.net/introduction.html |title=Python Psyco Homepage at sourceforge |accessdate=2009-03-04 }}</ref><ref>{{cite web |url=http://www.scipy.org/PerformancePython |title=A beginners guide to using Python for performance computing at scipy.org |accessdate=2009-03-04 |url-status=dead |archiveurl=https://web.archive.org/web/20090311070246/http://www.scipy.org/PerformancePython |archivedate=2009-03-11 }}</ref><ref>{{cite web
|url=http://www.ibm.com/developerworks/library/l-psyco.html
|title=Charming Python: Make Python run as fast as C with Psyco
|website=[[IBM]]
|accessdate=2009-03-04 }}</ref>
The average speed improvement is typically in the 1.5-4x range, making Python performance close to languages such as [[Smalltalk]] and [[Scheme (programming language)|Scheme]], but still slower than compiled languages such as [[Fortran]], [[C (programming language)|C]] or some other [[just-in-time compiler|JIT]] languages like [[C Sharp (programming language)|C#]] and [[Java (programming language)|Java]].<ref>{{cite web |url=http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=all&d=data&gcc=on&gpp=on&java=on&csharp=on&psyco=on&mzscheme=on&vw=on&python=on&calc=calculate&box=1 |title=Boxplot Summary |accessdate=2009-10-16 |url-status=dead |archiveurl=https://web.archive.org/web/20110603231929/http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=all&d=data&gcc=on&gpp=on&java=on&csharp=on&psyco=on&mzscheme=on&vw=on&python=on&calc=calculate&box=1 |archivedate=2011-06-03 }}</ref>
Psyco also advertises its ease of use: the simplest Psyco optimization involves adding only two lines to the top of a script:<ref>{{cite web|last=Rigo|first=Armin|title=Quick examples|url=http://psyco.sourceforge.net/psycoguide/node8.html|work=The Ultimate Psyco Guide|accessdate=3 June 2011}}</ref>
<syntaxhighlight lang="python">
import psyco
psyco.full()
</syntaxhighlight>
These commands will import the psyco module, and have Psyco optimize the entire script. This approach is best suited to shorter scripts, but demonstrates the minimal amount of work needed to begin applying Psyco optimizations to an existing program.
==See also==
{{Portal|Free and open-source software}}
* [[PyPy]]
* [[Unladen Swallow]]
* [[Cython]]
* [[YARV]] (Yet another Ruby VM)
==References==
{{Reflist}}
==External links==
* {{sourceforge|psyco}}
* David Mertz's IBM developerWorks article: [http://www-106.ibm.com/developerworks/linux/library/l-psyco.html Make Python run as fast as C with Psyco]
* [https://archive.today/20121217175529/https://www.pooryorick.com/secure/wiki/Pub/Psyco psyco notes, Poor Yorick]
{{Python (programming language)}}
[[Category:Free software programmed in Python]]
[[Category:Python (programming language) implementations]]
[[Category:Python (programming language) software]]
[[Category:Software using the MIT license]]
{{Free-software-stub}}