If you make an array of substr references to the characters in a buffer, and then overlay each line into that buffer, the cost of performing the splitting/indexing of the strings is done once:

#! perl -slw
use strict;

my $c = $ARGV[ 0 ] // 25;

my $buf = chr(0)  x 62;
my @cRefs = map \substr( $buf, $_, 1 ), 0 .. length( $buf )-1;

until( eof( DATA ) ) {
    substr( $buf, 0 ) = <DATA>;
    print ${ $cRefs[ $c ] };
}

__DATA__
ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789abcdefghijklmnopqrstuvwxyz
ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789abcdefghijklmnopqrstuvwxyz
ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789abcdefghijklmnopqrstuvwxyz
ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789abcdefghijklmnopqrstuvwxyz
[download]

A few runs:

C:\test>1202693 0
A
A
A
A
C:\test>1202693 25
Z
Z
Z
Z
C:\test>1202693 32
6
6
6
6
C:\test>1202693 61
z
z
z
z
[download]