All the differences in timing results presented here (at least for while (), while (1), and for(;;)) are statistically insignificant. The proof for this statement comes from investigating what code is being run in each case. This can be done by using the Perl bytecode compiler (B::Bytecode).

If you execute the following:

perl -MO=Bytecode,-H,-ofor_loop    -e 'for (;;) {}'
perl -MO=Bytecode,-H,-owhile_loop  -e 'while () {}'
perl -MO=Bytecode,-H,-owhile1_loop -e 'while (1) {}'
[download]

and then diff the resulting files (for_loop, while_loop and while1_loop), you will find that they are all completely identical. This means that all three are exactly equivalent as far as code execution goes. Therefore, which of the three idioms to use is just a matter of personal preference.