US 12,094,030 B2
Method and apparatus for runing a service, and electronic device
Zhichao Li, Los Angeles, CA (US); Sikai Qi, Beijing (CN); Zherui Liu, Beijing (CN); Yibo Zhu, Los Angeles, CA (US); Chuanxiong Guo, Los Angeles, CA (US); Cheng Tan, Los Angeles, CA (US); Jian Zhang, Los Angeles, CA (US); and Jian Wang, Beijing (CN)
Assigned to Lemon Inc., Grand Cayman (KY)
Filed by Lemon Inc., Grand Cayman (KY)
Filed on Dec. 7, 2023, as Appl. No. 18/532,819.
Application 18/532,819 is a continuation of application No. PCT/SG2022/050601, filed on Aug. 23, 2022.
Claims priority of application No. 202111088174.4 (CN), filed on Sep. 16, 2021.
Prior Publication US 2024/0104687 A1, Mar. 28, 2024
Int. Cl. G06T 1/20 (2006.01)
CPC G06T 1/20 (2013.01) 19 Claims
OG exemplary drawing
 
1. A method for running a service, comprising:
determining a target deployment manner of a graphics processing unit (GPU) according to performance data of each service in a service set, wherein the deployment manner comprises: dividing the GPU into sub-GPUs of a respective size and determining a service configured to be run by each sub-GPU; and
switching, for the service in the service set, running of the service from a sub-GPU indicated by a current deployment manner to a sub-GPU indicated by the target deployment manner, the switching comprising:
executing a switching step for each service:
determining, for the service, a first sub-GPU to be newly created and a second sub-GPU to be deleted by comparing the current deployment manner with the target deployment manner of the GPU;
pairing the first sub-GPU with the second sub-GPU for the service and obtaining at least one sub-GPU pair on the premise that a target performance of the service is satisfied;
creating the first sub-GPU comprised in the sub-GPU pair, running the service on the created first sub-GPU and then deleting the second sub-GPU comprised in the sub-GPU pair for each sub-GPU pair; and
creating a first sub-GPU, running the service on the created first sub-GPU and then deleting a second sub-GPU, for a first sub-GPU and a second sub-GPU that are not paired for the service.