No abstract available.

This paper reports on a new bottom-up synthesis technique for general behavioral descriptions. Our technique extends traditional straight-line code synthesis by allowing hierarchical, block-structured dataflow graphs with block-level parallelism. Program path probabilities are taken into account; and both high-level synthesis and design-space exploration are addressed.

A novel hierarchical top down placement technique is presented for circuits implemented in the sea-of-gates design style. It is based on a new hypergraph multi-partitioning algorithm, whose time complexity is linear in the number of pins of a circuit. The partitioning algorithm uses Steiner tress for the modeling of net topologies, which allows taking wiring congestion into account during placement. This leads to a more sophisticated balance criterion compared to conventional min-cut algorithms and consequently to a better distribution of active elements and wiring over the chip area. Experimental results show that the application of the new method makes the wiring of designs considerably easier.>

A novel heuristic for global routing in graphs is developed. Based on a flow model it can handle many nets simultaneously, thus reducing the net ordering problem. To demonstrate the validity of the method it was applied to standard cell design style. For this application the authors combined the flow model approach with linear assignment to achieve a hierarchical global routing scheme. This procedure is about six times faster than the old flat net by net global router, while producing denser layouts for the majority of testcases. Good quality was achieved in comparison with the TimberwolfSC version 5.4 global router.>

The problem stated and solved in the above paper1has been solved before with essentially the same algorithm by Hashimoto and Stevens.

A new data structure suitable for the implementation of gridless routing algorithms is presented. The data structure and some basic access procedures are outlined in PASCAL and experimental results are given to demonstrate the performance of retrieval algorithms working on the structure.