A direct sysread call seems to be >2x faster:

$ ls -sh bigfile.txt
2.8G bigfile.txt

$ cat direct_sysread.pl

use Fcntl qw(:DEFAULT);
use Symbol;

my $file = shift @ARGV;

my $fh = gensym;
sysopen $fh, $file, O_RDONLY;
my $content;
sysread $fh, $content, -s $file;

$ cat file_slurp.pl

use File::Slurp;

my $file = shift @ARGV;
open my $fh, "<", $file or die $!;

my $content = read_file ($file);

$ time direct_sysread.pl bigfile.txt

real    0m1.785s
user    0m0.008s
sys     0m1.776s

$ time file_slurp.pl bigfile.txt

real    0m4.994s
user    0m0.924s
sys     0m4.052s
[download]