Most LLVM projects I see are simply replacing -O3 pipelines. i.e. the code has already been heavily stress-tested prior to a safer optimization port of a stripped binary.

I also do time critical stuff, so llvm is a nonstarter for predictive latency in code motion. For most other use-cases, clang/llvm typically does improve performance a bit, and I do like its behavior on ARM.

Happy coding =)