ARM Cortex-A15 Compiler Optimizations

Obviously as the compiler applies more optimizations, the build time increases.

-Ofast yielded the fastest performance for the C-Ray ray-tracing. The GCC optimization levels improved each step of the way for this multi-threaded ray-tracer.

The -O3 optimization flag was acting strange for the Cortex-A15 with GCC 4.7.2 on this test where its performance was only a little better than -O0. The -Ofast flag was nearly five times faster than GCC without any compiler optimizations.

The Nero2D performance improved each step of the way as expected for computationally demanding tests.

In March, we'll be back with the more interesting ARM Linux benchmarks.