LLVM source-to-source compiler for loop parallelization


I am a research scholar of Dept. of Computer Science and Engg. of Indian Institute of Technology Kharagpur.
I just want to know whether LLVM has any source-to-source compiler which automatically parallelizes a loop ?


To my knowledge, not within LLVM.

Within my group we developed pet, a polyhedral extraction tool [1]. pet is based on clang and used to parallelize loops in the latest version of Pluto [2] (CPU parallelism) as well as in ppcg (both GPU and CPU parallelism).


[1] http://impact.gforge.inria.fr/impact2012/workshop_IMPACT/verdoolaege.pdf

[2] http://pluto-compiler.sourceforge.net/
     A practical automatic polyhedral parallelizer and locality optimizer
     Bondhugula, Uday and Hartono, Albert and Ramanujam, J. and
     Sadayappan, P.
[3] git://repo.or.cz/ppcg.git
     Polyhedral parallel code generation for CUDA
     S Verdoolaege, J Carlos Juega, A Cohen