Conversion of circuit description to an abstract model of the circuit
View Patent ↗A system and method is disclosed for converting an existing circuit description from a lower level description, such as RTL, to a higher-level description, such as TLM, while raising the abstraction level. By changing the abstraction level, the conversion is not simply a code conversion from one language to another, but a process of learning the circuit using neural networks and representing the circuit using a system of equations that approximate the circuit behavior, particularly with respect to timing aspects. A higher level of abstraction eliminates much of the particular implementation details, and allows easier and faster design exploration, analysis, and test, before implementation. In one aspect, a model description of the circuit, protocol information relating to the circuit, and simulation data associated with the lower level description of the circuit are used to generate an abstract model of the circuit that approximates the circuit behavior.
1. A method for converting a description of a circuit into an abstract model, comprising:
using a computer to read a model description of the circuit, protocol information relating to the circuit, and simulation data associated with the circuit; and
based on the model description, the protocol information, and the simulation data, generating an abstract model of the circuit that approximates the circuit behavior, wherein generating an abstract model comprises:
generating a system of weighted equations representing the behavior of the circuit;
applying input patterns to the system of equations to generate actual output values;
calculating an error by using a difference between the actual values and desired values; and
modifying the weighted system of equations based on the calculated error; and
wherein generating an abstract model further comprises performing causality analysis, wherein performing causality analysis comprises defining a set of causality characters; and
reducing the set of causality characters through statistical analysis.
2. The method of claim 1 , wherein performing causality analysis comprises determining repetitive behavior between multiple input and output signals of the simulation data.
3. The method of claim 1 , further comprising further reducing the set of causality characters using a genetic optimization algorithm.
4. The method of claim 1 , wherein generating the abstract model further includes generating fork tables and latency tables wherein the fork tables include state information and path information associated with the circuit and the latency tables include a latency period associated with a simulated event.
5. The method of claim 1 , wherein the simulation data is at a transaction level.
6. The method of claim 1 , wherein generating an abstract model includes learning the circuit using neural networks.
7. The method of claim 1 , wherein the simulation data is at a transaction-level through a conversion including:
reading low-level simulation data obtained from simulation of the circuit description;
analyzing switching signals, from the low-level simulation data, on simulated hardware lines of the circuit;
determining transactions of the protocol associated with the analyzed switching signals; and
storing a converted description of the simulated circuit as a series of the determined transactions.
8. The method of claim 1 , wherein the circuit description is in a register transfer level and the abstract model is at a transaction-level model, the transaction-level model having a different level of abstraction than the register transfer level.
9. The method of claim 1 , wherein at least one of the model description of the circuit, the protocol information relating to the circuit, or the simulation data, is provided from a client computer coupled to a network and generating the abstract model is performed on a server computer coupled to the network.
10. A storage device including instructions stored thereon for performing a method for converting a description of a circuit into an abstract model, the method comprising:
using a computer to read a model description of the circuit, protocol information relating to the circuit, and simulation data associated with the circuit; and
based on the model description, the protocol information, and the simulation data, generating an abstract model of the circuit that approximates the circuit behavior, wherein generating an abstract model includes performing causality analysis, the causality analysis including analyzing inputs and outputs to the circuit to find a repetitive correlation therebetween wherein generating an abstract model further includes learning the circuit using neural networks, wherein learning comprises:
generating a system of weighted equations representing the behavior of the circuit;
applying input patterns to the system of equations to generate actual output values;
calculating an error using a difference between the actual values and desired values; and
modifying the weightings in the system of equations based on the calculated error.
11. The storage device of claim 10 , wherein generating the abstract model further includes generating a system of equations that represent the behavior of the circuit.
12. The storage device of claim 10 , wherein
performing causality analysis further includes defining a set of causality characters, each causality character representing an association between an event and an output; and
wherein the method further comprises reducing the set of causality characters through statistical analysis.
13. The storage device of claim 12 , further reducing the set of causality characters using a genetic optimization algorithm.
14. The storage device of claim 10 , wherein generating the abstract model further includes generating fork tables and latency tables wherein the fork tables include state information and path information associated with the circuit and the latency tables include a latency period associated with a simulated event.
15. The storage device of claim 10 , wherein the simulation data is at a transaction level.
16. The storage device of claim 10 , wherein the simulation data is at a transaction-level through a conversion including:
reading low-level simulation data obtained from simulation of the circuit description;
analyzing switching signals, from the low-level simulation data, on simulated hardware lines of the circuit;
determining transactions of the protocol associated with the analyzed switching signals; and
storing a converted description of the simulated circuit as a series of the determined transactions.
17. The storage device of claim 10 , wherein the circuit description is in a register transfer level and the abstract model is at a transaction-level model, the transaction-level model having a different level of abstraction than the register transfer level.
18. An apparatus for converting a circuit description into an abstract model of the circuit, comprising:
means for reading a model description of the circuit, protocol information relating to the circuit, and simulation data associated with the circuit; and
means for generating an abstract model of the circuit that substantially imitates the circuit behavior using the model description of the circuit, the protocol information, and the simulation data, wherein generating an abstract model comprises:
generating weighted equations representing the behavior of the circuit;
applying input patterns to the weighted equations to generate output values;
calculating an error based on the output values and desired values; and
modifying the weighted equations based on the calculated error; and
wherein generating an abstract model further comprises generating fork tables and latency tables wherein the fork tables include state information and path information associated with the circuit and the latency tables include a latency period associated with a simulated event.