Dear Monks, I am having fun with timestamps

I am extracting a lot of data from a data warehouse and have to convert the legacy timestamps (see code) to human readable. Of course the requirement has changed and I now need epoch time too. This code runs on every row returned for a lot of data so I want to be a little efficient.

Any optimisations of the following code, or other ways to do it welcome.

use strict; use warnings;
use POSIX qw(mktime);

for my $t (<DATA>) {
  # Handle time, this is stored as a 'Candle' timestamp
  # "CYYMMDDhhmmssSSS" where C is the century (0 for 20th, 1 for 21st,
+ etc). 
  chomp ($t);
  print $t;
  my @bits = split //, $t; # use these later
  
  # This is how I convert it to human readable now
  my $c=substr $t,1,1,''; # grab the century
  substr $t, 2,0,'-';substr $t, 5,0,'-';substr $t, 8,0,' ';   # do the
+ date
  substr $t,11,0,':';substr $t,14,0,':';substr $t,17,0,'.'; # do the t
+ime
  $t = 19+$c.$t; # add back the century
  print " = $t\n";

  # This is what I am thinking to go to epoch seconds
  # of course I would grab my human version from here too
  my $C = $bits[0] + 19;
  my $Y = $C . (join "", @bits[ 1,2]);
  my $M = join "", @bits[ 3,4 ];
  my $D = join "", @bits[ 5,6 ];
  my $h = join "", @bits[ 7,8 ];
  my $m = join "", @bits[ 9,10];
  my $s = join "", @bits[11,12];
  print "c:$C y:$Y M:$M D:$D h:$h m:$m s:$s\n";
  my $epoch = mktime($s, $m, $h, $D, $M-1, $Y-1900,);
  print scalar localtime $epoch;
  print $/;
}
__DATA__
1130508154533613
1130508160033800
1130508161534113
1130508163034519
1130508164535019
1130508164535019
1130508170035191
1130508171533160
1130508173033660
1130508174534097
1130508180034535
1130424070116695
1130425070118790
1130426070118834
1130427070122888
1130428070123115
1130429070126161
1130430070127538
1130501070128195
1130502070131210
1130503070131969
1130504070135198
[download]

Cheers,
R.