OpenMP 3.1 Support Readied By Intel For LLVM Clang
One of the major limitations of LLVM/Clang has been its lack of OpenMP support though there's been numerous developers to attempt implementing support for this multi-processing API. While many have tried, no full OpenMP support has yet to land in upstream LLVM and Clang.
The full OpenMP 3.1 code by Intel for Clang is hopeful that it will be upstreamed, but for now it's on a separate code branch and is being maintained against the Clang 3.3 stable release. Some of the initial patches for this support have been merged into trunk.
Andrey Bokhanko of Intel says that this OpenMP 3.1 implementation in Clang passes all tests and the performance is said to be comparable to other OpenMP-enabled compilers. While the patches are now coming out of Intel, the work was originally started at AMD.
More information on this full OpenMP 3.1 support for Clang can be read about in this mailing list message while the code for now is hosted on GitHub.