Method and system for analyzing satisfiability of transient pulse reconvergence phenomenon of combined circuit

Abstract

The invention discloses a method and system for analyzing satisfiability of the transient pulse reconvergence phenomenon of a combined circuit. The method comprises then following steps: a NAND graphformat file is read in, a start node V is randomly selected, a depth-first method is used to search and arbitrarily select an output node P reachable by V, the maximum distance and the minimum distance from any node on the path to the output node P are calculated, a to-be-checked list C is initialized, an element (V, V) is added to the list C to be checked, a sensitization constraint is added to each element in the list C to be checked, an SAT solver is used to determine whether there is an input vector that enables two paths, which can be sensitized at the same time, between the starting nodeV and the output node P. If not, the list C to be checked is updated, and the elements in the list C to be checked are sorted and re-solved by the SAT solver. The invention has the advantages of large analyzable scale, higher accuracy, small calculation scale, high calculation efficiency and short algorithm running time.

Description

technical field [0001] The invention relates to the technical field of circuit analysis of integrated circuits, in particular to a method and system for analyzing the satisfiability of transient pulse reconvergence phenomena of combined circuits for fast and accurate analysis of the satisfiability of reconvergence phenomena of pulse signals. Background technique [0002] The phenomenon of reconvergence is very common in modern integrated circuit design. Some studies have pointed out that the analysis of some classic chips found that more than half of the units are fan-out units. Due to the limitation of chip output pins, a large number of re-convergence of fan-out branches will inevitably occur. The transient pulse generated by a certain unit will be delayed by multiple paths and may meet at a certain point. The reconverged pulse has great uncertainty in waveform, amplitude and pulse width. In addition, due to the existence of logic shielding, the transient pulse cannot pro...

AI Technical Summary

Problems solved by technology

The input vector of the simulation method is randomly generated, which may not cover all the input vectors that can cause reconvergence

The method of probabilistic analysis will have accuracy problems, and figure 1 For example, the probability that the output y is 1 is 37.5% (truth table), but it is calculated as 42.19% by the method of probability propagation, because the correlation between signals is not considered

The binary decision tree method can analyze the reconvergence phenomenon more accurately, but the solution scale is exponential, which is not applicable to large-scale circuits

Images