| 摘要 |
A distributed digital rule processor to create a single system image (SSI) on a cluster of von Neumann processors. The rule processing engine has a group of digital rule nets each having their own local bindery and local controllers which are interconnected by a global bindery. Slave translators are interfaced to the global bindery to receive rule calls from the rule nets and return statuses. A compiler maps the typical source code instructions into groups of rules which are stored in the rule nets' storage area. The rules contain groups of input variables and output variables. The output variables of a calling rule become the input and data for the called rule. Slave application processes on the von Neumann processors perform slave activities for the slave translators, as the rule nets are unable to manipulate data. Rules received from the rule nets by slave translators are processed and processed data or return statuses are sent back to the nets for further instructions. The rule nets contain the code compiled by the programmer and the application data is stored in the global buffers. This separates the code from data. The digital rules include node, processor, and thread addressing variables to permit extending the model to handle multiple nodes in a Wide Area Network, Local Area Network, System Area Network, or cluster.
|
