I figure you still have some indenting work to do. This is very important. Judiciously applied white space is one of the very most important things that you can do to improve readability of your code.

Untested, but I figure close to what you want...test, experiment, move forward with the advice you've gotten so far...

my $num_errors = 0;
foreach my $file (keys %hash1)
{
   if (!exists ($hash2{$file}) )
   {
       print "file: $file doesn't exist in 2nd directory\n";
   }
   elsif ($hash1{$file} ne $hash2{$file})
   {
       print "md5 didn't match for $file\n";
          # meaning that file in 2nd directory is not the 
          # same as the file in 1st directory
       $num_errors++;
   }
}
print "total errors = $num_errors\n";
[download]

Perl "sees" something akin to this (below)... a bit harder to understand than the above?
white space is important, variable names are important.
I called my hashes %dirA and %dirB instead of %hash1 and %hash2 for a reason!

%x is a hash but "x" has no contextual meaning! %dirA is a hash of file names in directory A to checksums.
Even %dirA_files_to_checksums would be wayyyyyy better than %hash1. I guess %dir1 is also ok. The % means hash - give more contextual information!

my $num_errors=0;foreach my $file (keys %hash1){if (!exists ($hash2{$f
+ile})){print "file: $file doesn't exist in 2nd directory\n";}elsif ($
+hash1{$file} ne $hash2{$file}){print "md5 didn't match for $file\n";$
+num_errors++;}}print "total errors = $num_errors\n";
[download]

PS: Yes, a hash tables for this purpose is going to be WAY more efficient than an array.