I just joined your group to find out more about LLVM.
I've been reading the documents and am confused at a higher level
I installed version 3.1 per the documentation and after a try or two,
was successful in generating a working 'clang', etc.
My interest is (1) to see if I can get improved runtime performance for Perl5
and (2) for the future to build a Perl5 front-end compiler using LLVM.
I was able to generate a working Perl5.16.0 with '-O2' that passed all tests and then
tested against some real code. All worked flawlessly!
Performance was approximately the same as 'GCC'. I then compiled with '-O3'
and was pleasantly surprised to find about 15% improvement on 'cpu' intensive code.
That's more than I expected.
I've been reading 'http://llvm.org/docs/tutorial/LangImpl4.html’ and I followed
the directions for '...enhanced with the LLVM JIT and optimizer', but without any
So is this as far as I can go with-out writing a front-end, or are there steps that I
can take to get more optimization in the existing Perl5? Any pointers are appreciated.