Back to my question, as I understand, perl code also need to be compiled and excuted(like C).

That's not exactly how it works. The way perlcc worked was to write a C program that contained the Perl interpreter as well as the source code in a serialized format. The program launched the Perl interpreter, then fed it the source code. If you think about what it takes to run Perl, you'll realize that you need the whole interpreter around in case you have an eval STRING construct of some kind. There's no direct translation of Perl statements to assembly language, which I think is what you wanted.

The whole point is moot though because perlcc never worked very well and no one wants to maintain it.