#1618: Compile with adaptive MPI (based on Charm++) -------------------------------+-------------------------------------------- Reporter: jtao@… | Owner: Type: enhancement | Status: new Priority: major | Milestone: Component: Cactus | Version: development version Resolution: | Keywords: -------------------------------+--------------------------------------------
Comment (by eschnett):
The reason Jian says that NR may not benefit is that simulation with large number of ghost zones may not benefit. Switching to other numerical methods, such as DG, may change the picture.
Re Carpet: Most of Carpet's global variables live in a namespace. The reason this was done was one of convenience -- it is easier to access a namespace than a singleton class object where the pointer has to be retrieved from cctkGH. The changes to avoid global variables are straightforward, but tedious.
There exist other global variables having to do with profiling, timing, keeping statistics etc. Those should probably be put into thread-local storage.