What gave you that impression?

my $f;
open($f,'<',"bigfile") or die "Could not open file\n";

my $i;
while ($i++<1000000) {
  seek($f,10,0);
  seek($f,3000,0);
}
[download]

This script needs 1.2 seconds on my machine, for 2 millions seeks. Time for the loop without anything in it is 0.275 seconds, so we are at 460 nanoseconds per seek

my $f;
open($f,'<',"bigfile") or die "Could not open file\n";

my $i;
my $g;
while ($i++<1000000) {
  seek($f,10,0);
  $g= <$f>;
  seek($f,3000,0);
}

print length $g,"\n";
[download]

This script additionally does one million reads of a 2 character string and runs 3 seconds. That means 1800 nanoseconds is the minimal time for a read (from cache obviously, if the data is on disk multiply that time by 1000 or more)

Actually I'm of the impression that seek function may eat up a lot of time if I'm suppose to execute it in a loop.

I need to traverse a file and I may require constant vertical movement. But then performance is an issue here.

I'm willing to initialise file pointers at the place where it is required. Is there a way to do that. Thanks for your reply.