Hello, I 've a question about the use of MPI_F_SYNC_REG to avoid optimisation problem. If I understand well a way to avoid  change in order is to use MPI_F_SYNC_REG after the wait like this :   call MPI_ISEND(buf,...,req)   ! some work   call MPI_WAIT(req,...)   if (.not. MPI_ASYNC_PROTECTS_NONBLOCKING) call MPI_F_SYNC_REG(buf)   buf = val2 My question is what in this solution prevent the compiler to put the "call MPI_WAIT" after the line "buf = val2" ? Regards