Cache-only memory architecture

Cache only memory architecture is first introduced by E. Hagersten, A. Landin, and S. Haridi, in "DDM-A cache-only memory architecture" on IEEE Computer 1992. As the name suggested, this arthitecture is composed by caches only as Hagersten found that CC-NUMA machines have a set of local copies of data in cache and also have the same data in a distant memory, which can be eliminated using this method. In this method, what is kept is the cache which does not have a permanent home, since there is no memory to fall back on.