US 11,853,662 B2
Machine-learning enhanced compiler
Sankaranarayanan Srinivasan, San Jose, CA (US); Senthilkumar Thoravi Rajavel, Hillsboro, OR (US); Vinod Kumar Nakkala, Sunnyvale, CA (US); Avinash Anantharamu, Sunnyvale, CA (US); Pierre Clement, Antony (FR); Saibal Ghosh, Sunnyvale, CA (US); Sashikala Oblisetty, Cupertino, CA (US); and Etienne Lepercq, Shrewsbury, MA (US)
Assigned to Synopsys, Inc., Sunnyvale, CA (US)
Filed by Synopsys, Inc., Mountain View, CA (US)
Filed on Jun. 21, 2022, as Appl. No. 17/845,421.
Application 17/845,421 is a continuation of application No. 16/992,636, filed on Aug. 13, 2020, granted, now 11,366,948.
Claims priority of application No. 19204152 (EP), filed on Oct. 18, 2019.
Prior Publication US 2022/0318468 A1, Oct. 6, 2022
Int. Cl. G06F 30/27 (2020.01); G06F 30/3323 (2020.01); G06F 119/12 (2020.01)
CPC G06F 30/27 (2020.01) [G06F 30/3323 (2020.01); G06F 2119/12 (2020.01)] 17 Claims
OG exemplary drawing
 
1. A method comprising:
storing a base model generated using base data;
receiving training data generated by compiling circuit designs;
generating, using the training data, a tuned model;
generating, using the training data and the base data, a hybrid model, wherein the base model, the tuned model, and the hybrid model effects of place and route strategies;
receiving a cost function selected from a plurality of cost functions comprising a first cost function that biases towards reducing compilation time, a second cost function that biases towards reducing resource consumption during compilation, and a third cost function that biases towards both reducing compilation time and reducing resource consumption during compilation; and
biasing the base model, the tuned model, and the hybrid model using the selected cost function.