I'm very green to the compiler development scene.
Is there a specific way to measure the effectiveness of a compiler's optimizations, or would I simply compare execution times between direct compilation and optimized output?
This is not an open ended question, I'm not looking for somebody to count off QA software on their fingers or give me a list of wikipedia links. I want to know if I'm missing one big obvious thing that everyone else knows from having experience.
The most common performance metrics for compiler optimization is execution time for a defined set of examples, or the size when comparing size-optimizations. Since a typical optimization often works better for one type of code and worse for another, the challenge is to define this set of examples, or benchmarks, to be as close to typical code as possible.