A Highly Parallel VLSI-Based Subsystem of the NON-VON Database Machine

Shaw, David Elliot; Ibrahim, Hussein; Andrews, Jim; Wiederhold, Gio

The NON·VON machine (portions of which are presently under construction in the Department of Computer Science at Columbia, in cooperation with the Knowledge Base Management Systems Project at Stanford) was designed to apply computational parallelism on a rather massive scale to a large share or the information processing functions now performed by digital computers. The NON·VON architecture comprises a tree-structured Primary Processing Subsystem (PPS), which we are implementing using custom nMOS VLSI chips, and a Secondary Processing Subsystem (SPS) incorporating modified highly intelligent disk drives. NON-VON should permit particularly dramatic performance improvements in very large scale data manipulation tasks, including relational database operations and external sorting. This paper includes a brief overview of the NON-VON project and a more detailed discussion of the structure and function or the PPS unit and its constituent processing elements.


More About This Work

Academic Units
Computer Science
Department of Computer Science, Columbia University
Columbia University Computer Science Technical Reports, CUCS-017-81
Published Here
October 18, 2011