I recently did a poor man's benchmark just for kicks. The results might at least be interesting to you.

I wrote this little script to populate a db with 50,000 rows into a db table of three columns (one just an autoinc id).

#!/usr/bin/perl

use strict;
use warnings;

use Data::Dumper; $Data::Dumper::Indent = 3;
use Digest::MD5 qw/md5_base64/;

use base qw/DBIx::Class::Schema::Loader/;
DBIx::Class::Schema::Loader->loader_options( relationships => 1 );

my $schema = DBIx::Class::Schema::Loader->connect('dbi:SQLite:/tmp/tes
+t');
#my $schema = DBIx::Class::Schema::Loader->connect('dbi:Pg:dbname=test
+', 'postgres', '' );
#my $schema = DBIx::Class::Schema::Loader->connect('dbi:mysql:dbname=t
+est', 'mysql', '' );


my @inserts;

for ( 1 .. 50000 ) {

    push @inserts,
        {
          foo => md5_base64( $_ . 'foo' ),
          bar => md5_base64( $_ . 'bar' )
        };

}

my $new_row = $schema->resultset('TestTable');

$new_row->populate( \@inserts );
[download]

Loading to postgresql-server-8.1.9-2.1 took just under 8 minutes, mysql-5.0.26-12 about 35 seconds, sqlite-3.3.8-14 even with synchronous mode turned off couldn't even seem to handle 1000 rows a minute (I didn't wait for it to finish all 50,000 rows). I reused this script just now for writing 50,000 lines to a flat text file and it was done almost before I pressed enter.

Obviously YMMV. My workstation is a sun ultra 40 dual opteron with a pair of sata drives that I striped into a raid 0 and I'm running opensuse 10.2.