One process doesnot support more than 100 threads?

That limitation is due to the stupidly large default stack size allocated to each thread. You can overcome that by using the stack_size => 4096 detailed in the module's POD.

This runs 500 threads, uses around 1.5 GB and heads 1000 urls in 24 seconds via my 2Mbs connection:

#! perl -slw
use strict;
use LWP::Simple;
use threads stack_size => 4096;
use threads::shared;
use Thread::Queue;

our $THREADS //= 500;

my %log :shared;
my $Q = new Thread::Queue;

my @threads = map async( sub {
    while( my $url = $Q->dequeue() ) {
        my @info = head 'http://' . $url;
        lock %log;
        $log{ $url } = join $;, map{ local $^W; $_ // '*n/a*' } @info;
    }
} ), 1 .. $THREADS;

while( <> ) {
    chomp;
    sleep 1 if $Q->pending > $THREADS;
    $Q->enqueue( $_ );
}

$Q->enqueue( (undef) x $THREADS );

$_->join for @threads;

my( $url, $status );
print "$url : $status" while ( $url, $status ) = each %log;

__END__
c:\test>t-head-urls -THREADS=100 urls.list.small
www.t-mobile.com : text/html; charset=utf-8?109511?873075813?*n/a*?Mic
+rosoft-IIS/7.0
www.avocent.com : text/html; charset=UTF-8?114030?*n/a*?*n/a*?Microsof
+t-IIS/7.0
www.fsw.com : text/html?9926?1306121940?*n/a*?Apache/2.2.21 (Unix) mod
+_ssl/2.2.21 OpenSSL/0.9.7a mod_auth_passthrough/2.1 mod_bwlimited/1.4
+ FrontPage/5.0.2.2635 mod_fcgid/2.3.5 Resin/3.1.10 Sun-ONE-ASP/4.0.3
www.voodoopc.com : text/html?6021?*n/a*?1328712575?Apache
www.creative.com : text/html; Charset=iso-8859-1?42330?*n/a*?132870531
+8?Microsoft-IIS/6.0
www.belkin.com :
www.argus-systems.com : text/html?8111?1320697393?*n/a*?Apache/2.2.6 (
+Unix) mod_python/3.3.1 Python/2.5.1 mod_ssl/2.2.6 OpenSSL/0.9.8g
www.iss.net : text/html; charset=UTF-8?*n/a*?*n/a*?*n/a*?Apache
www.Joyent.COM : text/html; charset=UTF-8?*n/a*?*n/a*?*n/a*?Apache
...
[download]

Creating more (Perl) threads than you have CPUs / cores in your machine makes little sense. Use an asynchronous fetch library, like WWW::Curl and/or AnyEvent to fetch multiple URLs within one child process or thread, instead of spawning a separate thread or child process for each URL.