Abstract

Redundant execution schemes for high-assurance processors have historically enforced cycle-accurate determinism across two identical cores, a constraint that can disable out-of-order execution and inflate silicon area.  This publication describes a verification method in which two heterogeneous processing cores deposit execution checkpoints into a shared hardware monitor peripheral.  Each checkpoint is an N-bit token (e.g., a 32-bit token in an exemplary implementation) carrying a validity bit, a checkpoint type bit, a rolling sequence number, and a payload value.  The peripheral buffers each token stream in a hardware queue, compares the two queue heads, and raises a sticky, non-maskable alarm on a mismatch or a timeout.  The method thereby preserves diagnostic coverage across independent clock frequencies and dissimilar pipelines.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS