[Beowulf] Book recomendations for cluster programing

[Micha Feigin]

I'm trying to learn cluster/parallel programing properly. I've got some
information on MPI, although I'm not sure if it's the best books. I was
wondering if you have some book recommendations regarding the more specialized
things, especially the cpu vs gpu paralelization issue (or as far as I
understood, data intensive vs memory intensive programming and how to
convert programs from one to the other) proper program/cluster topology
considerations and such.

Thanks