Here's a little example that avoids the overhead of manipulating @buffer and @rows (I wrote it as an example program without the database stuff.)

open(FH, 'data.txt');
my $c_size = 500;
while (!eof FH) {
  for (map {(eof) ? () : [split ',', <FH>]} 1..$c_size) {
    print "Insert @$_\n";
  }
  print "Buffer done\n";
}
[download]