#928: openmp parallelization within Exact broken ------------------------------------+--------------------------------------- Reporter: knarf | Owner: Type: defect | Status: confirmed Priority: minor | Milestone: ET_2012_11 Component: EinsteinToolkit thorn | Version: development version Resolution: | Keywords: ------------------------------------+---------------------------------------
Comment (by eschnett):
"single" alone would not solve the issue. It ensures that only a single thread executes the code (similar to "critical"), but would still be executed for every grid point for this thread. To ensure that the other threads see the result of the single region, one would need some kind of synchronisation, probably either a barrier or (again) an if statement checking first_call. I don't see any advantage. Could you give a particular example?