Date: Sat, 20 Apr 2002 11:03:54 +1100 (edt) From: Andrew MacIntyre <andymac@bullseye.apana.org.au> To: rob <rob@pythonemproject.com> Cc: Alexander Leidinger <Alexander@Leidinger.net>, <perky@fallin.lv>, <ports@FreeBSD.ORG> Subject: Re: icc-compiled Python Message-ID: <Pine.OS2.4.32.0204201051500.16970-100000@tenring.andymac.org> In-Reply-To: <3CC0201C.E32A0CA2@pythonemproject.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 19 Apr 2002, rob wrote:
> Sorry I missed the beginning of the thread. Is this Icc Python faster
> than the normal one? I am wondering as I've found FreeBSD Python 2.1 to
> be 1/2 the speed of the ActiveState Windows version on Win2k, using
> Numeric Python. This was using an FDTD benchmark and a Romberg variable
> width integration benchmark. Rob.
The info missing from the original post was the gcc optimisation options,
not to mention the icc options.
Just using ./configure yields "-O3". Adding a "-march=<cpu_family>"
setting can have noticeable impact on performance - on my P5-166SMP box
"-march=pentium" improved overall python performance by ~10%.
On the subject of Numeric, I've seen material that suggests gcc 2.8.1 is
capable of producing better fp code than 2.9.x (no info about gcc 3+). My
own experience on OS/2 also suggests that gcc 2.8.1 "-O2" beats gcc 2.9.2
"-O<anything>" for python performance.
--
Andrew I MacIntyre "These thoughts are mine alone..."
E-mail: andymac@bullseye.apana.org.au | Snail: PO Box 370
andymac@pcug.org.au | Belconnen ACT 2616
Web: http://www.andymac.org/ | Australia
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.OS2.4.32.0204201051500.16970-100000>