Xputer

From Wikipedia, the free encyclopedia

The data-stream-based Xputer machine paradigm is the counterpart of the instruction-stream-based von Neumann machine paradigm. A von Neumann computer has a program counter. An Xputer, however, has one or several data counters, instead. The GAG methodology is an important enabling technology to implement data sequencing by data counters on Xputers and for Reconfigurable Computing. The Xputer paradigm, also called "anti machine" is derived from a generalization of the systolic array and the date stream definitions coming along with it: the Super systolic array, also called Kress/Kung machine. The dichotomy of (von Neumann) "computer" and Xputer (anti machine) explains the paradigm shift from classical computing to Reconfigurable Computing. First Xputer architectures have been developed and implemented in the 1980ies by Reiner Hartenstein and his team at the University of Kaiserslautern, where speed-ups by factors up to 15,000 have been obtained by software to configware migration Because FPGAs at that time have been very small, a programmable PLA called DPLA has been used for Xputer architectures. The DPLA has been designed and fabricated in the framework of the Mead-&-Conway style German E.I.S. multi university project.

The term "xputer" (non-von-Neumann) should not be confused with the "transputer" (von Neumann).

Prior to both transputers and Xputers comes UK Patent GB 2069196B on Self Configuring Cellular Multiprocessor Computer + corresponding US and EEC patents.[1] Conceived in 1979 by Bernard de Neumann, this massively parallel self-configuring multi-processor computer (ie with unconstrained architecture), and patent applied for in 1980. The patent was awarded in 1984.

[edit] See also

[edit] External links