Recent work has developed support for compiler-based performance tools specific to analysis and transformation of large scale programs. This talk will present how we approach global analysis through a merged representation of the AST across files to simplify how tools can work on large scale applications which would ordinarily be prohibitively expensive (taking many hours) to have a tool compile. The recent work reduces the time required by several ordered of magnitude and permits analysis tools to be developed more simply. This talk will also present recent OpenMP auto-parallelization work that supports how existing parallel MPI software can support hybrid MPI/OpenMP. This work is part of continuing research on OpenMP and programming models within the ROSE project and part of the ROSE-OpenMP compiler.