There are some who have proposed renaming the chapter
to either "1.5-sided" or "2-sided-1-sided". :-)
-] In fact, we don't need to relax the 1-sided model to make it possible to
-] implement on the shared-memory systems. The shared-memory vendors have a
-] straightforwad way to implement the current model by using the mirrored buffer
-] technique for window memory that was not alocated with RMA_MALLOC. It requires
-] duplicating the window in shared memory and then using WINDOW_IN and WINDOW_OUT
-] to copy data to and from shared memory buffer in order to enforce consistency. Of
-] course, the copy operation degrades performance and extra memory is required, but
-] it should work.
I agree. It should work but it makes many machines that
could make optimal use of shared memory slower.
Most machines require that shared memory be allocated by the
OS (these machines include almost all popular [mass-production]
UNIX boxes as well as other OSs). There are some "specialty"
machines that don't have this requirement but I think these
machines are in the vast minority in numbers.
I feel that the relaxation of RMA_MALLOC has come about
entirely for FORTRAN convenience. However, the Forum
has voted.