The performance implication depends on your implementation. If it's a critical application, particularly one that iterates significantly, the difference may be significant. Conversely, if the code isn't iterative, or only sees a few iterations, the difference is likely neglible. Given this understanding, you may want to benchmark the codes using the Benchmark module:

use Benchmark qw(cmpthese);

@data = (1,2,3,4); # the values that would have been in @_

my $minCPUsec = 15;

cmpthese(-$minCPUsec, {
  with =>    \&sub_with_ifs,
  without => \&sub_without_ifs
});
[download]

And change @_ in the referenced subs to @data.

Additionally, you may be interested in the Benchmarking Your Code tutorial, but I'd recommend using cmpthese instead of timethese.