17887968. APPARATUS AND METHOD WITH SCHEDULING simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

From WikiPatents
Jump to navigation Jump to search

APPARATUS AND METHOD WITH SCHEDULING

Organization Name

SAMSUNG ELECTRONICS CO., LTD.

Inventor(s)

Jae Wook Lee of Seoul (KR)

Younghwan Oh of Yongin-si (KR)

Yunho Jin of Seoul (KR)

Tae Jun Ham of Seoul (KR)

APPARATUS AND METHOD WITH SCHEDULING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17887968 titled 'APPARATUS AND METHOD WITH SCHEDULING

Simplified Explanation

The abstract describes a method for scheduling the execution of multiple models in an accelerator based on quality of service (QoS) information and idle time. Here is a simplified explanation of the abstract:

  • The method involves receiving execution requests for multiple models that are executed independently in an accelerator.
  • For each model, the method predicts quality of service (QoS) information, which corresponds to the expected performance or user experience of the model.
  • The method then schedules the execution of the models in units of layers (components) of the models.
  • The scheduling decision is based on either the QoS information, the idle time (time when the accelerator is not busy), or both.
  • The scheduling aims to optimize the execution of the models by considering their QoS requirements and the availability of resources.

Potential applications of this technology:

  • This method can be applied in various fields where multiple models need to be executed in an accelerator, such as machine learning, data analytics, and scientific simulations.
  • It can be used in cloud computing environments to efficiently schedule the execution of models on accelerators, improving resource utilization and user experience.
  • This technology can also be beneficial in edge computing scenarios, where models are executed on local accelerators to reduce latency and improve responsiveness.

Problems solved by this technology:

  • Efficiently scheduling the execution of multiple models in an accelerator can be challenging, especially when considering their different QoS requirements and the availability of resources.
  • This method addresses the problem of optimizing the scheduling decision by considering both QoS information and idle time.
  • By intelligently scheduling the models, it can help ensure that the accelerator is utilized effectively and that the models meet their QoS requirements.

Benefits of this technology:

  • Improved resource utilization: By considering idle time and QoS information, the method can optimize the scheduling of models, leading to better utilization of the accelerator's resources.
  • Enhanced user experience: By prioritizing models based on their QoS requirements, the method can improve the overall user experience by ensuring that models with higher QoS demands are executed in a timely manner.
  • Increased efficiency: The method's ability to predict QoS information and schedule models accordingly can lead to more efficient execution, reducing overall execution time and improving system performance.


Original Abstract Submitted

A processor-implemented method with scheduling includes: receiving one or more execution requests for a plurality of models executed independently of each other in an accelerator; predicting, for each of the plurality of models, quality of service (QoS) information corresponding to the model; and scheduling the plurality of models in units of layers of the plurality of models based on, for each of the plurality of models, either one or both of the QoS information and an idle time occurring in response to a candidate layer to be scheduled in the model being executed in the accelerator.

Retrieved from "http://wikipatents.org/index.php?title=17887968._APPARATUS_AND_METHOD_WITH_SCHEDULING_simplified_abstract_(SAMSUNG_ELECTRONICS_CO.,_LTD.)&oldid=22005"