Temporal Tracing of On-chip Signals using Timeprints

Rehab Massoud, Hoang M. Le, Peter Chini, Prakash Saivasan, Roland Meyer, Rolf Drechsler

In: Design Automation Conference (DAC). Design Automation Conference (DAC-2019) June 2-6 Las Vegas Nevada United States 2019.


This paper introduces a new method to tracecycle-accuratelythe temporal behavior of on-chip signals while operating in-field.Current cycle-accurate schemes incur unacceptable amounts ofdata for logging, storage and processing.Our key idea to enable efficient yet cycle-accurate tracing, isto bringtimingto the front as a main traced artifact. We split thesignal tracing into consecutive (back-to-back) finite trace-cycles.Within a trace-cycle, a signal’s value-change instance gets assignedan encoded timestamp. At the end of each trace-cycle, these en-coded timestamps are aggregated into a loggedtimeprint, whichsummarizes the temporal behavior over the trace-cycle.To retrieve the accurate timing, we reconstruct the exact in-stances from a timeprint via a SAT query. The experiments demon-strate how unprecedented lightweight tracing can be applied, andhow timeprints enable the verification of cycle-accurate propertiesand the detection of sporadic temperature effects


German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz