US 12,449,889 B2
DVFS controlling method, semiconductor device and semiconductor system using the DVFS controlling method
Jung Ho Kim, Suwon-si (KR); and Sung Sup Lim, Yongin-si (KR)
Assigned to Samsung Electronics Co., Ltd.
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed on Jan. 13, 2023, as Appl. No. 18/096,798.
Claims priority of application No. 10-2022-0021309 (KR), filed on Feb. 18, 2022.
Prior Publication US 2023/0266815 A1, Aug. 24, 2023
Int. Cl. G06F 1/3296 (2019.01); G06F 9/48 (2006.01); G06F 11/34 (2006.01)
CPC G06F 1/3296 (2013.01) [G06F 9/4881 (2013.01); G06F 11/3419 (2013.01)] 20 Claims
OG exemplary drawing
 
16. A DVFS control method for determining an operating frequency of a processor executing a plurality of neural network models, the DVFS control method comprising:
receiving data characterizing each neural network model of the plurality of neural network models;
based on the data characterizing each neural network model of the plurality of neural network models, determining an execution time of the processor for each neural network model of the plurality of neural network models and a workload of the processor for each neural network model of the plurality of neural network models;
determining a total workload of the processor based on the workload of the processor for each neural network model of the plurality of neural network models; and
controlling the operating frequency of the processor based on the total workload of the processor.