Machine-learning based datapath extraction
View Patent ↗A datapath extraction tool uses machine-learning models to selectively classify clusters of cells in an integrated circuit design as either datapath logic or non-datapath logic based on cluster features. A support vector machine and a neural network can be used to build compact and run-time efficient models. A cluster is classified as datapath if both the support vector machine and the neural network indicate that it is datapath-like. The cluster features may include automorphism generators for the cell clusters, or physical information based on the cell locations from a previous (e.g., global) placement, such as a ratio of a total cell area for a given cluster to a half-perimeter of a bounding box for the given cluster.
1. A computer-implemented method of training a machine-learning based datapath extraction tool, comprising:
receiving a learning set of design patterns representing clusters of interconnected cells wherein each design pattern is designated as either datapath or non-datapath, by executing first instructions in a computer system;
identifying cluster features in the design patterns, by executing second instructions in the computer system;
applying one or more machine-learning models to the cluster features to selectively associate the features with either datapath logic or non-datapath logic, by executing third instructions in the computer system; and
calibrating the one or more machine-learning models, by executing fourth instructions in the computer system.
2. The method of claim 1 , further comprising:
applying the one or more machine-learning models to a validation set of design patterns representing different clusters of interconnected cells to classify the design patterns in the validation set as either datapath or non-datapath;
determining that the one or more machine-learning models did not accurately classify the design patterns in the validation set; and
re-calibrating the one or more machine-learning models responsive to said determining.
3. The method of claim 1 wherein said identifying cluster features includes computing one or more automorphism generators for the cell clusters.
4. The method of claim 1 wherein said applying applies at least two machine-learning models each providing an indication of whether a given one of the cell clusters is datapath logic, and the given cell cluster is classified as datapath only when both of the two machine-learning models indicate that the given cell cluster is datapath logic.
5. The method of claim 4 wherein a first one of the machine-learning models is a support vector machine, and a second one of the machine-learning models is a neural network.
6. The method of claim 5 wherein said calibrating includes:
adjusting a separation threshold used by the support vector machine for a hyperplane boundary between datapath logic support vectors and non-datapath logic support vectors; and
adjusting a numeric score of the neural network.
7. A computer system comprising:
one or more processors which process program instructions;
a memory device connected to said one or more processors; and
program instructions residing in said memory device for training a machine-learning based datapath extraction tool by receiving a learning set of design patterns representing clusters of interconnected cells wherein each design pattern is designated as either datapath or non-datapath, identifying cluster features in the design patterns, applying one or more machine-learning models to the cluster features to selectively associate the features with either datapath logic or non-datapath logic, and calibrating the one or more machine-learning models.
8. The computer system of claim 7 wherein said program instructions further apply the one or more machine-learning models to a validation set of design patterns representing different clusters of interconnected cells to classify the design patterns in the validation set as either datapath or non-datapath, determine that the one or more machine-learning models did not accurately classify the design patterns in the validation set, and re-calibrate the one or more machine-learning models responsive to said determining.
9. The computer system of claim 7 wherein the identifying of the cluster features includes computing one or more automorphism generators for the cell clusters.
10. The computer system of claim 7 wherein the applying applies at least two machine-learning models each providing an indication of whether a given one of the cell clusters is datapath logic, and the given cell cluster is classified as datapath only when both of the two machine-learning models indicate that the given cell cluster is datapath logic.
11. The computer system of claim 10 wherein a first one of the machine-learning models is a support vector machine, and a second one of the machine-learning models is a neural network.
12. The computer system of claim 11 wherein the calibrating includes:
adjusting a separation threshold used by the support vector machine for a hyperplane boundary between datapath logic support vectors and non-datapath logic support vectors; and
adjusting a numeric score of the neural network.
13. A computer program product comprising:
a computer-readable storage medium; and
program instructions residing in said storage medium for training a machine-learning based datapath extraction tool by receiving a learning set of design patterns representing clusters of interconnected cells wherein each design pattern is designated as either datapath or non-datapath, identifying cluster features in the design patterns, applying one or more machine-learning models to the cluster features to selectively associate the features with either datapath logic or non-datapath logic, and calibrating the one or more machine-learning models.
14. The computer program product of claim 13 wherein said program instructions further apply the one or more machine-learning models to a validation set of design patterns representing different clusters of interconnected cells to classify the design patterns in the validation set as either datapath or non-datapath, determine that the one or more machine-learning models did not accurately classify the design patterns in the validation set, and re-calibrate the one or more machine-learning models responsive to said determining.
15. The computer program product of claim 13 wherein the identifying of the cluster features includes computing one or more automorphism generators for the cell clusters.
16. The computer program product of claim 13 wherein the applying applies at least two machine-learning models each providing an indication of whether a given one of the cell clusters is datapath logic, and the given cell cluster is classified as datapath only when both of the two machine-learning models indicate that the given cell cluster is datapath logic.
17. The computer program product of claim 16 wherein a first one of the machine-learning models is a support vector machine, and a second one of the machine-learning models is a neural network.
18. The computer program product of claim 17 wherein the calibrating includes:
adjusting a separation threshold used by the support vector machine for a hyperplane boundary between datapath logic support vectors and non-datapath logic support vectors; and
adjusting a numeric score of the neural network.