I’m running test with SWMR. Does parallel HDF5 supports SWMR? For example, I run my simulation code with multiple MPI processes and each of these processes writes a block of a 2D global array. At each simulation step, a new 2D global array is generated and it needs to be appended to the previous one. So basically, the simulation output is a 3D array with an unlimited dimension that can grow during runtime. If parallel HDF5 supports SWMR, is there an example code I can look at?