develooper Front page | perl.perl5.porters | Postings from January 2001

Tweaked config for Solaris/sparc == 30% perfomance boost

Thread Next
From:
Alan Burlison
Date:
January 9, 2001 08:17
Subject:
Tweaked config for Solaris/sparc == 30% perfomance boost
Message ID:
3A5B399B.F6736627@uk.sun.com
Some timings based on my app which does intensive object
allocation/deallocation, both built with Forte C workshop 6.1:

'Standard' perl configuration (5.6.0)
-------------------------------------
config_args='-dsOE -Dprefix=/home1/software/perl/debug
-Dinstallprefix=/home1/software/perl/debug
-Dsiteprefix=/home1/software/perl/debug -Ubincompat5005 -D
use64bitint'
optimize='-O'

real    0m32.07s
user    0m25.61s
sys     0m6.19s

Optimised perl configuration (5.6.0)
------------------------------------
config_args='-dsOE -Dprefix=/home1/software/perl/debug
-Dinstallprefix=/home1/software/perl/debug
-Dsiteprefix=/home1/software/perl/debug -Ubincompat5005 -D
use64bitint -Uusemymalloc -Ud_sigsetjmp -Doptimize=-xO4 -xdepend'
optimize='-xO4 -xdepend'

real    0m22.32s
user    0m22.11s
sys     0m0.14s

That is a 30% performance improvement, without changing a single line of
code.  This is the main reason I intend ditching siglongjmp when I
integrate 5.6.1 into Solaris.

Interestingly enough, the difference between the 'make test' results is
only about 15%.  I suspect this is because the test suite is a series of
relatively short 'straight line' code that doesn't involve lots of loops
or object creation.

u=0.56  s=1.28  cu=36.79  cs=26.76  scripts=230  tests=10119
u=0.54  s=0.91  cu=43.28  cs=26.90  scripts=230  tests=10119

The moral of the story is the test suite is not a good benchmark.

Alan Burlison

Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About