Warning: All code untested

Your code for keys is incomplete, because it does not tell you which keys in %old_gp are not in %new_gp. Starting from blazar's suggestion, I would expand:

my @diff = grep { !exists $old_gp{$_} } keys %new_gp;
push @diff, grep { !exists $new_gp{$_} } keys %old_gp;
[download]

More generally:

my (%foo, %bar);
# Populate %foo and %bar

# Keys present in only one hash
my @only_in_foo = grep { ! exists $bar{$_} } keys %foo;
my @only_in_bar = grep { ! exists $foo{$_} } keys %bar;

# Matching keys which have differences in values
my @kv_different = map {
   if (exists $bar{$_} && $bar{$_} ne $foo{$_}) { $_ }
   else { () }
} keys %foo;

# Matching keys and values
my @kv_equal = map {
   if (exists $bar{$_} && $bar{$_} eq $foo{$_}) { $_ }
   else { () }
} keys %foo;
[download]

More compactly (but not necessarily more efficient or readable, it's a matter of taste):

my (@only_in_foo, @kv_different, @kv_equal);
foreach (keys %foo) {
   if (exists $bar{$_}) {
      if ($bar{$_} eq $foo{$_}) {
         push @kv_equal, $_ 
      }
      else {
         push @kv_different, $_ 
      }
   }
   else {
      push @only_in_foo, $_ 
   }
}
my @only_in_bar = grep { ! exists $foo{$_} } keys %bar;
[download]