Technique for determining circuit interdependencies
Embodiments of a device (such as a computer system or a circuit tester), a method, and a computer-program product (i.e., software) for use with the device are described. These systems and processes may be used to statistically characterize interdependencies between sub-circuits in an integrated circuit (which are referred to as ‘aggressor-victim relationships’). In particular, statistical relationships between the aggressors and victims are determined from values of a performance metric (such as clock speed) when the integrated circuit fails for a group of state-change difference vectors. Using these statistical relationships, a worst-case sub-group of the state-change difference vectors, such as the worst-case sub-group, is selected. This sub-group can be used to accurately test the integrated circuit.
1. A method for selecting a sub-group of state-change difference vectors from a group of state-change difference vectors for testing a circuit, comprising:
using a computer to generate the group state-change difference vectors, wherein a given state-change difference vector corresponds to a change in a state of circuit components in the circuit, and wherein one or more elements in the given state-change difference vector specify state transitions for one or more of the circuit components;
determining values of a performance metric for the group of state-change difference vectors when the circuit fails during testing, wherein a given value of the performance metric corresponds to the given state-change difference vector;
for a given sub-circuit in a set of sub-circuits in the circuit and the group of state-change difference vectors, calculating statistical relationships between a transition type in the given sub-circuit when the circuit failed during testing and transition types in other sub-circuits in the set of sub-circuits, wherein a statistical relationship is determined for each pairing of the given sub-circuit with one of the other sub-circuits;
repeating the calculating for all of the sub-circuits in the set of sub-circuits; and
selecting the sub-group of state-change difference vectors based on the calculated statistical relationships, wherein the selected state-change difference vectors are those that result in failure of the circuit during testing for a range of values of the performance metric.
2. The method of claim 1 , wherein the group of state-change difference vectors is generated using a pseudo-random generator.
3. The method of claim 1 , wherein a number of state-change difference vectors in the group of state-change difference vectors is based on a number of logic levels and fan-in between gates in the circuit.
4. The method of claim 1 , wherein the circuit components include logic circuits.
5. The method of claim 1 , wherein the circuit components include gates.
6. The method of claim 1 , wherein the performance metric includes clock speed.
7. The method of claim 1 , wherein the performance metric includes power consumption.
8. The method of claim 1 , wherein determining the values of the performance metric involves measurements on a current version of the circuit.
9. The method of claim 1 , wherein the transition types in the other sub-circuits include a transition from a first logical state to a second logical state, a transition from the second logical state to the first logical state, and no transition.
10. The method of claim 1 , wherein the transition types in the other sub-circuits can include the transition type in the given sub-circuit.
11. The method of claim 1 , further comprising selecting the set of sub-circuits in the circuit prior to calculating the statistical relationships.
12. The method of claim 1 , wherein the statistical relationship for each pairing corresponds to co-occurrences of the transition type in the given sub-circuit and one of the transition types in the one of the other sub-circuits for multiple state-change difference vectors in the group of state-change difference vectors.
13. The method of claim 12 , wherein a co-occurrence for a given state-change difference vector includes a presence of the transition type in the given sub-circuit and a presence of the one of the transition types in the one of the other sub-circuits, or an absence of the transition type in the given sub-circuit and an absence of the one of the transition types in the one of the other sub-circuits.
14. The method of claim 1 , wherein the statistical relationship for each pairing corresponds to cross-occurrences of the transition type in the given sub-circuit and one of the transition types in the one of the other sub-circuits for multiple state-change difference vectors in the group of state-change difference vectors.
15. The method of claim 14 , wherein a cross-occurrence for a given state-change difference vector includes a presence of the transition type in the given sub-circuit and an absence of the one of the transition types in the one of the other sub-circuits, or an absence of the transition type in the given sub-circuit and a presence of the one of the transition types in the one of the other sub-circuits.
16. The method of claim 1 , further comprising determining a margin for the performance metric using the selected sub-group of state-change difference vectors.
17. The method of claim 16 , further comprising revising the circuit based on the determined margin or changing manufacturing of the circuit based on the determined margin.
18. The method of claim 1 , wherein failure of the circuit for the given state-change difference vector is associated with cross-talk.
19. A computer-program product for use in conjunction with a computer system, the computer-program product comprising a non-transitory computer-readable storage medium and a computer-program mechanism embedded therein for selecting a sub-group of state-change difference vectors from a group of state-change difference vectors for testing a circuit, the computer-program mechanism including:
instructions for generating the group state-change difference vectors, wherein a given state-change difference vector corresponds to a change in a state of circuit components in the circuit, and wherein one or more elements in the given state-change difference vector specify state transitions for one or more of the circuit components;
instructions for determining values of a performance metric for the group of state-change difference vectors when the circuit fails during testing, wherein a given value of the performance metric corresponds to the given state-change difference vector;
for a given sub-circuit in a set of sub-circuits in the circuit and the group of state-change difference vectors, instructions for calculating statistical relationships between a transition type in the given sub-circuit when the circuit failed during testing and transition types in other sub-circuits in the set of sub-circuits, wherein a statistical relationship is determined for each pairing of the given sub-circuit with one of the other sub-circuits;
instructions for repeating the calculating for all of the sub-circuits in the set of sub-circuits; and
instructions for selecting the sub-group of state-change difference vectors based on the calculated statistical relationships, wherein the selected state-change difference vectors are those that result in failure of the circuit during testing for a range of values of the performance metric.
20. A computer system, comprising:
a processor;
memory;
a program module, wherein the program module is stored in the memory and configured to be executed by the processor, wherein the program module selects a sub-group of state-change difference vectors from a group of state-change difference vectors for testing a circuit, the program module including:
instructions for generating the group state-change difference vectors, wherein a given state-change difference vector corresponds to a change in a state of circuit components in the circuit, and wherein one or more elements in the given state-change difference vector specify state transitions for one or more of the circuit components;
instructions for determining values of a performance metric for the group of state-change difference vectors when the circuit fails during testing, wherein a given value of the performance metric corresponds to the given state-change difference vector;
for a given sub-circuit in a set of sub-circuits in the circuit and the group of state-change difference vectors, instructions for calculating statistical relationships between a transition type in the given sub-circuit when the circuit failed during testing and transition types in other sub-circuits in the set of sub-circuits, wherein a statistical relationship is determined for each pairing of the given sub-circuit with one of the other sub-circuits;
instructions for repeating the calculating for all of the sub-circuits in the set of sub-circuits; and
instructions for selecting the sub-group of state-change difference vectors based on the calculated statistical relationships, wherein the selected state-change difference vectors are those that result in failure of the circuit during testing for a range of values of the performance metric.