Artificial Intelligence
Optimal composition of real-time systems
Artificial Intelligence
Path planning under time-dependent uncertainty
UAI'95 Proceedings of the Eleventh conference on Uncertainty in artificial intelligence
| Hi-index | 0.00 |
High-performance CAD systems help designers of VLSI logic to synthesize optimal circuits. With these, users are able to repeatedly select and invoke the most appropriate-looking command (which corresponds to an algorithm). They are hard for novice users to operate, however, because efficient command-selection requires experience. In the past, novices used a "command script with heuristics" for selecting commands and automating command invocation, but the heuristics needed to be rewritten for any updated version of the CAD system being used, and users were unable to place their own deadlines on the time within which the design results had to be obtained. To cope with these problems, we have developed an Intelligent Command Control Shell (ICCS) which performs logic synthesis tasks by automatically selecting and executing multiple sequences of commands within a pre-set time limit or until it obtains an adequate circuit. ICCS uses easily up datable statistical data as its knowledge base. ICCS also features "time-constrained control", which takes imposed deadlines into account in its selection of commands, so as to produce the best possible circuit within a given time limit. When applied to the design of large-scale practical circuits, the use of ICCS resulted in circuits with 6% shorter delay on average (and 30% shorter delay in the best case) than those obtained with simple optimization command.