Abstract:
A machine learning-based process includes identifying a first set of features that includes features of a reference implementation of a circuit design and features of a synthesized version of a modified version of the circuit design. A first classification model is applied to the first set of features, and the first classification model indicates a full implementation flow or an incremental implementation flow. The full implementation flow is performed on the synthesized version of the modified version in response to the first classification model indicating the full implementation flow, and the incremental implementation flow is performed on the synthesized version of the modified version in response to the first classification model indicating the incremental implementation flow. The full and incremental implementation flows generate implementation data that is suitable for making an integrated circuit (IC).
Abstract:
Processing a circuit design includes determining that an operating frequency for a first placement and routing for the circuit design does not exceed a target operating frequency, distinguishing between loop paths and feed-forward paths in the circuit design, and, responsive to determining that the operating frequency does not exceed the target operating frequency, relaxing timing constraints of the feed-forward paths using a processor. A second placement and routing is performed on the loop paths and the feed-forward paths of the circuit design.