Re^3: Fast, Efficient Union and Intersection on arrays

[admiral_grinder]

I remember one time getting a noticeable speed improvement by using the 'for (...;...;...)' over a foreach loop for a large array. I wondering if that would have helped here.

[TGI]

Once again, expectations are shattered: the c style for loop is the slowest.

sub one_cfor {
    # my (@a, @b);
        # initialize @a and @b here

        my %hash;
        @hash{@a} = (1) x @a;

        my @union = @a;
        my @intersection;
        for (my $i=0; $i<$#b; $i++ ) {
            my $bval = $b[$i];
            if( exists $hash{$bval} ) { 
                push @intersection, $bval 
            }
            else {
                push @union, $bval 
            }
        }

#        print "union @union\n";
#        print "interstection @intersection\n";
}


################
Range (1,50) vs (3,80)
            Rate  one_cfor   one_for two_greps
one_cfor  4438/s        --      -24%      -30%
one_for   5834/s       31%        --       -8%
two_greps 6312/s       42%        8%        --
Range (1,5) vs (3,8)
             Rate  one_cfor   one_for two_greps
one_cfor  50000/s        --      -13%      -29%
one_for   57637/s       15%        --      -18%
two_greps 70323/s       41%       22%        --