System and method for adaptation of coherence models between agents

Abstract

A system and method are disclosed for multiple coherent caches supporting agents that use different, incompatible coherence models. Compatibility is implemented by translators that accept coherency requests and snoop responses from an agent and accept snoop requests and coherency responses from a coherence controller. The translators issue corresponding coherency requests and snoop responses to the coherence controller and issue corresponding coherency responses and snoop requests to the agent. Interaction between translators and the coherence controller accord with a generic coherence model, which may be a subset, superset, or partially inclusive of features of any native coherence model. A generic coherence protocol may include binary values for each of characteristics: valid or invalid, owned or non-owned, unique or shared, and clean or dirty.

Description

CROSS REFERENCE TO RELATED APPLICATIONS[0001]This application is related to and a continuation-in-part of U.S. Non-Provisional Utility patent application Ser. No. 14 / 806,786 titled DISTRIBUTED IMPLEMENTATION FOR CACHE COHERENCE filed on Jul. 23, 2015 by FORREST, Craig Stephen et al., the entire disclosure of which is incorporated herein by reference.FIELD OF THE INVENTION[0002]The invention is in the field of system-on-chip (SoC) and, more specifically, heterogeneous processing units, including processor cores, graphics processing units, and other components that manipulate or move data.BACKGROUND[0003]Modern systems-on-chip (SoCs) are often designed with heterogeneous processing units that are selected for their different system characteristics. Typically, a processing unit, or agent, implements a cache for storing a local copy of data, and that agent assumes a particular coherence model, which defines a set of cache states and a set of coherent transactions that effect transitions...

AI Technical Summary

Benefits of technology

The invention describes a translator that is aware of the allowed states and behaviors of a coherence model on its interfaces. The translator ensures that the destination provides the necessary capabilities for the source. The translator makes the coherence model of agents and coherence controllers clear to each other so that each only needs to be designed according to its coherence protocol. As a result, a correctly functioning system can be implemented. In simpler terms, the invention allows for transparent communication between agents and controllers.

Problems solved by technology

However, different processing units, delivered as intellectual property blocks from different vendors, may implement different and incompatible coherence models, making integration of those processing units in a SoC difficult if not impossible.

As a result, a SoC designer is faced with either higher design costs and longer development cycles or selection of compatible, but less optimal, processing units.

Images