Unless you are doing this thousands of times, it will hardly matter which one is faster.

If you must know, you can always use the Benchmark module, and run a simple test:

#!/usr/bin/perl

use strict;
use warnings;

use Benchmark 'cmpthese';

our $item = '"path/path1==new';

cmpthese(-1, {
    method1  =>  'my ($epath, $relt) =  split /==/, $item',
    method2  =>  'my ($epath, $relt) = $item =~ /([^=]*)==(.*)/',
});

__END__
            Rate method2 method1
method2 334042/s      --    -38%
method1 535924/s     60%      --
[download]

Method 1 will about 1.1 microseconds faster (at least, on my box). So, if I have a million of them, method1 will gain 1 second over method2. That's hardly worth the trouble.

I'd pick method1 because, IMO, it's more clear to see what it does.

Oh, and both methods use a regular expression anyway.