摘要:
A method and computer program product for partitioning the columns of a matrix A. The method includes providing the matrix A in a memory device of a computer system. The matrix A has n columns and m rows, wherein n is an integer of at least 3, and wherein m is an integer of at least 1. The method further includes executing an algorithm by a processor of the computer system. Executing the algorithm includes partitioning the n columns of the matrix A into a closed group of p clusters, wherein p is a positive integer of at least 2 and less than n, wherein the partitioning includes an affinity-based merging of clusters of the matrix A, and wherein each cluster is a collection of one or more columns of A.
摘要:
A computer system for partitioning the columns of a matrix A. The computer system includes a processor and a memory unit coupled to the processor. Program code in the memory unit, when executed by the processor, implements the method. Matrix A is provided in a memory device and has n columns and m rows; wherein n is an integer of at least 3; and wherein m is an integer of at least 1. The n columns is partitioned into a closed group of p clusters, p being a positive integer of at least 2 and less than n. The partitioning includes an affinity -based merging of clusters of pairs of clusters of the matrix A based on an affinity between the clusters in each pair of clusters being merged. Each cluster consists of one or more columns of matrix A. The p clusters are stored in a computer-readable storage device.
摘要:
A method and system for decreasing processing time in multi-corner static timing analysis. In one embodiment, parameters are ordered in a parameter order by decreasing magnitude of impact on variability of timing. In one example, a decreasing parameter order is utilized to order slack cutoff values that are assigned across a parameter process space. In another example, a decreasing parameter order is utilized to perform a multi-corner timing analysis on one or more dependent parameters in an independent fashion.
摘要:
A method and system for decreasing processing time in multi-corner static timing analysis. In one embodiment, slack cutoff values are assigned across a parameter process space. For example, a slack cutoff value is assigned to each parameter in a process space by determining an estimated maximum slack change between a starting corner and any other corner in a corresponding process sub-space. In another embodiment, parameters are ordered in a parameter order by decreasing magnitude of impact on variability of timing.
摘要:
A method of timing closure for integrated circuit designs uses multiple timing runs which distribute the frequency of identified fails per timing corner (between starting timing corners and remaining timing corners) to maximize efficiency in timing analysis. More specifically, the method closes timing for a chosen set of starting timing corners, verifies the remaining timing corners are orthogonal to the starting timing corners, closes timing for the remaining timing corners using multi-corner analysis, and verifies that all timing corners have positive slack margin.
摘要:
A method for performing a parallel static timing analysis in which multiple processes independently update a timing graph without requiring communication through a central coordinator module. Local processing queues are used to reduce locking overhead without causing excessive load imbalance. A parallel analysis is conducted on a circuit design represented by a timing graph formed by a plurality of interconnected nodes, the method including: using a computer for creating a shared work queue of ready to process independent nodes; assigning the independent nodes from the work queue to at least two parallel computation processes, simultaneously performing node analysis computations thereof; and modifying the circuit design by updating values of the processed independent nodes obtained from the node analysis, the at least two parallel computation processes independently updating the shared work queue to process a new plurality of independent nodes.
摘要:
A method for performing a parallel static timing analysis in which multiple processes independently update a timing graph without requiring communication through a central coordinator module. Local processing queues are used to reduce locking overhead without causing excessive load imbalance. A parallel analysis is conducted on a circuit design represented by a timing graph formed by a plurality of interconnected nodes, the method including: using a computer for creating a shared work queue of ready to process independent nodes; assigning the independent nodes from the work queue to at least two parallel computation processes, simultaneously performing node analysis computations thereof; and modifying the circuit design by updating values of the processed independent nodes obtained from the node analysis, the at least two parallel computation processes independently updating the shared work queue to process a new plurality of independent nodes.
摘要:
A method and a system for conducting a static timing analysis on a circuit having a plurality of point-to-point delay constraints between two points of the circuit, in which two conservative and two optimistic user defined tests are derived for all types of the point-to-point delay constraints. The method shows that when a conservative test is performed without introducing any special tags, then it is found that the point-to-point constraint is satisfied. On the other hand, when the optimistic test fails without any special tags, it is determined that the point-to-point constraint is bound to fail if special tags are introduced, in which case, they are to be introduced only when an exact slack is desired. Finally, for anything in between, a real analysis with special tags or path tracing is required. Based on the topology of the graph, arrival time based tests may be tighter in some situations, while the required arrival time based tests, may be tighter in others.
摘要:
A method of removing pessimism in static timing analysis is described. Delays are expressed as a function of discrete parameter settings allowing for both local and global variation to be taken in to account. Based on a specified target slack, each failing timing test is examined to determine a consistent set of parameter settings which produces the worst possible slack. The analysis is performed on a path basis. By considering only parameters which are in common to a particular data/clock path-pair, the number of process combinations that need to be explored is reduced when compared to analyzing all combinations of the global parameter settings. Further, if parameters are separable and linear, worst-case variable assignments for a particular clock/data path pair can be computed in linear time by independently assigning each parameter value. In addition, if available, the incremental delay change with respect to each physically realizable process variable may be used to project the worst-case variable assignment on a per-path basis without the need for performing explicit corner enumeration.
摘要:
A method for partitioning the columns of a matrix A. The method includes providing the matrix A in a memory device of a computer system. The matrix A has n columns and m rows, wherein n is an integer of at least 3, and wherein m is an integer of at least 1. The method further includes executing an algorithm by a processor of the computer system. Executing the algorithm includes partitioning the n columns of the matrix A into a closed group of p clusters, wherein p is a positive integer of at least 2 and less than n, wherein the partitioning includes an affinity-based merging of clusters of the matrix A, and wherein each cluster is a collection of one or more columns of A.