CACR Research Home

CACR Home

Continuum Computer Architecture (CCA) Web Site

Project Overview
Contact Information
Publications

 

Continuum Computer Architecture (CCA) is an alternative paradigm to the classical "von Neumann model." CCA will effectively utilize the billion transistors per chip of the near future.

The fundamental concept underlying CCA is that a general global model of parallel computation may be synthesized through a structure of higly replicated primitive functional blocks that unify in a single fine-grain element the basic properties of logic operations, information storage and naming, and information transfer and routing. A semantic synergism among the ensemble of fontons exhibit an emergent system behavior capable of executing higly parallel computing programs. Data structure and instruction sequences are distributed across potentially large regions of the simultac adn exection is achieved by the relative motion of these structures.

IMPACT:

Revolutionize means of exploiting future advanced technologies:
billion transistor semiconductor chips
quantum dots
superconductor RSFQ
Dramatically increase delivered performance per chip
Greatly simplify parallel program development and compiler design
Accelerate execution of DoD mission critical applications involving dynamic irregular data structures such as SAR, ATR, and C3I


NEW IDEAS:

New computing element that merges arithmetic, storage, communications, and flow control into one small highly replicated module.
Non-von Neumann parallel computing model achieves general global execution as an emergent behavior of replicated synergistic elements.
Achieve massive bi-section memory bandwidth as well as very high logic to storage ratio to enable new operating point for billion transistor chips.
Automatic resource management and load balancing made possible by very low overhead for synchronization and data movement.
Provide the means for memory to execute data in a manner comparable to conventional processors executing instructions.