Publikation

Power Intent from Initial ESL Prototypes: Extracting Power Management Parameters

David Lemma, Mehran Goli, Daniel Große, Rolf Drechsler

In: Proceedings of IEEE NorCAS Conference. IEEE Nordic Circuits and Systems Conference (NorCAS-2018) October 30-31 Tallinn Estonia 2018.

Abstrakt

The increase in digital circuit complexity not only stems from sophisticated functionality, but also from power concerns. Power concerns are addressed via the realization of power intent (the specification for power). Unlike functional specifications, power intent is generally implicit within an initial ESL Prototype. For power intent to become explicit, it needs to be expressed in terms of Power Management parameters. These parameters are major indicators of the efforts involved in realizing the power intent. We introduce an automated method to extract two Power Management parameters (number of Control Signals and Power Modes) from ESL prototypes. These parameters are extracted in a two-step process. First, relevant structural and behavioral information of the prototype is retrieved and translated into an activity profile. Following this, an analysis is performed on the activity profile to extract the power management parameters. The effectiveness and efficiency of the method is demonstrated by its application on several ESL benchmarks.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence