Back to results
US 12,748,943 B2Granted
Average power estimation using graph neural networks
AssigneeNVIDIA Corporation
InventorsYanqing Zhang, Haoxing Ren, Brucek Khailany
Life of this patent
Estimating the term…
Abstract
A graph neural network for average power estimation of netlists is trained with register toggle rates over a power window from an RTL simulation and gate level netlists as input features. Combinational gate toggle rates are applied as labels. The trained graph neural network is then applied to infer combinational gate toggle rates over a different power window of interest and/or different netlist.
Claims12
A computer system comprising:
- at least one data processor;
- a machine memory;
- a machine-readable media comprising instructions that, when executed by the at least one data processor, forms in the machine memory a neural network comprising:
- a fully-connected input layer configured to transform a netlist, input port toggle rates, and register toggle rates for an electronic circuit into inputs to a graph neural network;
- the graph neural network configured to transform a graph of the netlist, the input port toggle rates, and the register toggle rates into first inferred gate toggle rates for combinatorial elements of the electronic circuit in a power window;
- and
- at least one fully-connected output layer configured to transform the first inferred gate toggle rates into second inferred gate toggle rates for the combinatorial elements of the electronic circuit.
4 more elements
Dependencies are inferred from the claim text.