Identifying sequential redundancies without search

Abstract

Abstract

Sequential redundancy identification using recursive learning

“…For RID, only techniques that are based on uncontrollability and unobservability analysis are reviewed. A detailed review of other RID techniques can be found in [14]. …”

“…Otherwise, faults that can be detected by (sequential) multiple path sensitization would be marked as unobservable. In [14], a sequential redundancy identification procedure, FIRES, was proposed. FIRES applies sequential uncontrollability/unobservability on every fanout stem and finds c-cycle redundant faults for which a conflicting value assignment on the stem is necessary for detection.…”

“…But for some circuits, because the size of BDD can be exponentially large, the memory requirement may still pose a problem. FUNI only identifies untestable faults; it does not need the validation procedure in FIRES [14]. FUNI first finds illegal states, then analyzes every illegal state to determine the faults whose detection requires the fault-free circuit to enter that state.…”

“…The presence of untestable and redundant faults complicates automatic test pattern generation (ATPG). Additionally, redundancies induce many other detrimental effects [14]. Redundancies increase chip area, power consumption, and often, propagation delays in the circuit.…”

Exploiting power-up delay for sequential optimization

“…Another notion of equivalence was suggested in [7] which was used for sequential resynthesis by redundancy removal in [8]. Their notion of redundancy results in replacements which are not 0-cycle delay-safe replacements [9]; it has not been shown that if their replacements are n-cycle delay-safe replacements, neither what this n might be. A sequential resynthesis technique relevant to our model is retiming and resynthesis [10].…”