Power estimation method for an integrated circuit using probability calculations

A method of estimating power required for an integrated circuit including a logical circuit constituted by connecting a plurality of logical gates, comprising the steps of:

• partitioning said logical circuit into logical module sets consisting of a plurality of logical gates;
• representing logical values of respective nodes belonging to respective logical modules as a logical function of input signals of said logical modules, and then calculating a probability quantity corresponding to nodes of the respective logical modules and representative of a signal probability and a switching probability;
• expanding into a series a difference P.sub.e -P.sub.a between a strict value P.sub.e of said probability quantity and a value P.sub.a calculated based on an assumption that all inputs of the respective logical modules are independent, with respect to said probability quantity corresponding to entire input signals of said integrated circuit;
• and
• estimating power required for said integrated circuit by first calculating an infinite number of terms in said series as a correction term and then employing a value obtained by adding said correction term to said value P.sub.a as an approximate value of said probability quantity.

A method of estimating power required for an integrated circuit by calculating a probability of level transition at respective nodes of said integrated circuit, comprising the steps of:

• determining a minimum difference in delay time between a plurality of signals acting as input signals to gates in said integrated circuit;
• classifying gate inputs into groups according to the determined minimum differences in delay time;
• calculating a first transition probability considering a correlation within each of the respective groups of said classified gate inputs and disregarding the correlation between different groups of said classified gate inputs, and then calculating a second transition probability concerning gate outputs by virtue of the first calculated transition probability concerning said gate inputs;
• assuming a Markov nature on a probability of signals at more than two nodes, where the nodes depend on an identical signal;
• and
• disregarding correlation of signals between these nodes within a predetermined maximum allowable error if a minimum difference in delay time between said signals at more than two nodes is sufficiently large, until said identical signal reaches these nodes.

A method of estimating power consumption in an integrated circuit, comprising the steps of:

• storing input signals into primary input terminals of the integrated circuit in a storing unit;
• partitioning the input terminals into an input pattern in time series with a time interval;
• calculating a probability quantity of the input pattern provided to said primary input terminals at every time interval, the probability quantity representing at least one of a signal probability that logical values of output nodes of respective logical gates are 1 and a switching probability that logical gates are changed from 1 to 0;
• storing the probability quantity calculated at every time interval in the storing unit;
• calculating a power consumption in every time interval in a time series fashion as a function of the probability quantity;
• storing the power consumption in every time interval in a time series pattern;
• and
• disregarding correlation of signals between more than two nodes which depend on an identical signal X if a probability concerning logical values of said identical signal X satisfies a condition P(X)P(X).ltoreq.Error, wherein Error is a predetermined maximum allowable error, P(X) is a probability that said identical signal X is a logical value 0, and P(X) is a probability that said identical signal X is a logical value 1.

A method of estimating power required for an integrated circuit, comprising the steps of:

• storing into a storing unit descriptions of at least one of a circuit structure and a circuit operation of said integrated circuit;
• prior to the storing step, converting the descriptions into at least one of an evaluation equation and a data structure required for power estimation;
• estimating power of said integrated circuit by
• partitioning input terminals of said integrated circuit into an input pattern in time series with a time interval;
• calculating a probability quantity of the input pattern provided to said primary input terminals at every time interval, the probability quantity representing at least one of a signal probability that logical values of said output nodes of respective logical gates are 1 and a switching probability that logical gates are changed from 1 to 0;
• calculating a power consumption in every time interval in a time series fashion as a function of the probability quantity;
• and
• utilizing said evaluation equation or said data structure several times to estimate power.