DragonEgg 3.0 Puts GCC & LLVM In One Bed

With running the popular multi-threaded C-Ray ray-tracing benchmark, when LLVM 3.0 was touching the code it always ended up being faster. LLVM/Clang 3.0 was slightly faster than GCC 4.6.2 while using DragonEgg pushed it even faster. The best performance was with DragonEgg 3.0 and the option of both the GCC/LLVM optimizers; this led to a 14% performance improvement over vanilla GCC 4.6.2.

The single-threaded POV-Ray benchmark with its stock compiler flags was also much faster when taking advantage of the LLVM technologies than vanilla GCC. LLVM/Clang 3.0 was 17% faster than GCC for this Intel Gulftown platform while the DragonEgg plug-in pushed up the performance even more. However, when using DragonEgg with the "-fplugin-arg-dragonegg-enable-gcc-optzns" option for maximum performance, the test failed.

Clang still struggles with the Smallpt test due to the compiler front-end lacking OpenMP support. Between GCC and GCC with DragonEgg, there was not much of a performance difference.