US 12,079,608 B2
Efficient optimization for neural network deployment and execution
Ashutosh Pandey, Irvine, CA (US); Kaiping Li, Randolph, NJ (US); and Vikram Kumar Ramanna, San Jose, CA (US)
Assigned to Cypress Semiconductor Corporation, San Jose, CA (US)
Filed by Cypress Semiconductor Corporation, San Jose, CA (US)
Filed on Oct. 28, 2021, as Appl. No. 17/513,679.
Claims priority of provisional application 63/160,072, filed on Mar. 12, 2021.
Prior Publication US 2022/0303176 A1, Sep. 22, 2022
Int. Cl. G06F 9/50 (2006.01); G06F 8/41 (2018.01)
CPC G06F 8/447 (2013.01) [G06F 9/5016 (2013.01); G06F 9/5038 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method to deploy a machine learning model (MLM), the method comprising:
obtaining configuration settings of a pre-trained MLM, wherein the configuration settings of the MLM comprise:
an information characterizing a flow of data associated with the MLM, and
parameters of computational operations associated with the MLM;
obtaining a hardware configuration of a target computing device; and
compiling, in view of the configuration settings of the MLM and the hardware configuration of the target computing device, an execution package configured to execute the MLM on the target computing device, wherein compiling the execution package comprises:
identifying a format of vectorized instructions of the processor of the target computing device; and
generating a source code implementing the MLM on the target computing device, in view of the identified format of the vectorized instructions.