[Date Prev][Date Next][Thread Prev][Thread Next] [Search] [Date Index] [Thread Index]

Re: [MacPerl] MacPERL Performace vs. Intel Class systems



Tim Rand <tim@stlouis.datapage.com> writes:
}I have noticed that MacPERL runs SIGNIFICANTLY slower than on Wintel and
}REALLY want to see if this can't be fixed.
}
}Rather than provide some very obscure benchmark program, here is a simple
}little test program that I have used to benchmark various platforms:
}
}########### PERLTEST.PL ###############
}open(OUTFILE, ">ptest.dat") || die "Can't open ptest.dat";
}print STDERR "Starting\n";
}$starttime = time;
}        for ($i = 1; $i <= 100000; $i++) {
}                print OUTFILE "$i\n";
}        }
}$stoptime = time;
}print STDERR ($stoptime - $starttime), " Seconds elapsed time\n";
}#######################################
}
}This file will create a new file (ptest.dat) of about 600K containing the
}numbers from 1 to 100000.
}
}Here are some benchmarks I have run:
}
}Machine      Elapsed     Comment
}Pentium      12 Sec      Linux
}SPARC IPC    68 seconds  (with other processes running)
}SPARC 2      37 seconds  (with other processes running)
}SPARC Ultra   3 seconds  140 Mhz UltraSPARC
}
}Pentium 60   19 seconds  DOS 6.2
}Pentium 90   15 seconds  DOS 6.2
}386/25      287 seconds  DOS 6.2 <grin>
}486/66       22 seconds  Win 95
}Pent Pro/200  6 seconds  Win 95
}
}Mac IIci    180 seconds  68030 @ 16 Mhz
}Quadra 800   80 Seconds  68040 @ 33 Mhz
}Mac 8500/150 18 Seconds  (MacPERL installed for PowerPC)
}
}I cannot understand why the 8500 isn't much faster than a Pentium/60 and
}why a Quadra 800 isn't closer to the same speed as a 486/66.

One single benchmark means nothing; a fairer comparison would be a variety
of benchmarks that exercise various pieces of Perl.  (I wonder if such a
beast already exists on CPAN?)  But if I had to guess, the most likely
reason is because of something that Matthias can do nothing about, the
known-to-be-lame Macintosh file system.  (This is probably also the reason
I notice degradation of performace during downloads on my 180 MHz PPC
Performa 6400 and never notice the difference on my 80 MHz Sparc 2.)  Your
benchmark does intensive writing to the disk.

In 6 or so months, we can all upgrade what machines we can to Rhapsody and
lose the file system bottleneck.  Until then, there's little to be done.


--------
Paul J. Schinder
NASA Goddard Space Flight Center
Code 693
Greenbelt, MD 20770
schinder@pjstoaster.pg.md.us


***** Want to unsubscribe from this list?
***** Send mail with body "unsubscribe" to mac-perl-request@iis.ee.ethz.ch