Task Scheduling Method, Apparatus, and Computing System

A computing system, comprising:

• a plurality of partition schedulers;
• and
• partitions comprising computing units of the computing system and a plurality of task queues of different priorities configured based on task criticalities, wherein each partition corresponds to one partition scheduler of the plurality of partition schedulers, wherein a task criticality is based on a latency tolerance of a computing unit for a task run by the computing unit,
• wherein each partition scheduler is configured to determine partition criticalities of the partitions, wherein the partition criticalities indicate the latency tolerance of the computing unit in a partition for a to-be-scheduled task;
• wherein a second partition scheduler corresponding to a second partition in the partitions is configured to move, when it is determined, based on the partition criticalities of the partitions, to initiate inter-partition task scheduling, a first task in a first task queue in a first partition of the partitions to one of the task queues in the second partition, to cause one of the computing units in the second partition to process the first task, and wherein the first task queue has a first priority;
• and
• wherein a first partition scheduler corresponding to the first partition is configured to move, when it is determined, based on the partition criticalities of the partitions, to initiate the inter-partition task scheduling, a second task in a second task queue in the second partition to one of the task queues in the first partition, to cause a first computing unit of the computing units and in the first partition to process the second task, and wherein the second task queue has a second priority that is different from the first priority.

A computer program product comprising computer-executable instructions that are stored on a non-transitory computer-readable storage medium and that, when executed by one or more processors of an apparatus, cause the apparatus to:

• determine partition criticalities of partitions comprising computing units of a computing system and a plurality of task queues of different priorities configured based on task criticalities, wherein a task criticality is based on a latency tolerance of a computing unit for a task run by the computing unit, wherein the partition criticalities indicate the latency tolerance of the computing unit in a partition for a to-be-scheduled task, and wherein each partition corresponds to one partition scheduler of a plurality of partition schedulers;
• determine, based on the partition criticalities, to initiate inter-partition task scheduling;
• move, by a second partition scheduler as part of the inter-partition task scheduling, a first task in a first task queue in a first partition of the partitions to one of the task queues in a second partition of the partitions, to cause a second computing unit of the computing units and in the second partition to process the first task, wherein the second partition scheduler corresponds to the second partition and is of the plurality of partition schedulers, and wherein the first task queue has a first priority;
• and
• move, by a first partition scheduler as part of the inter-partition task scheduling, a second task in a second task queue in the second partition to one of the task queues in the first partition, to cause a first computing unit of the computing units and in the first partition to process the second task, wherein the first partition scheduler corresponds to the first partition and is of the plurality of partition schedulers, and wherein the second task queue has a second priority that is different from the first priority.