Dynamic Task Scheduling in Multi-Processor Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multiple processor systems, the insertion of dynamically generated tasks into the scheduling pipeline requires intervention from the central processing unit (CPU), leading to latency due to communication delays between the multiple processor system and the CPU.
Innovation Solution
The system allows direct communication between processing units and the scheduling unit, enabling dynamic tasks to be inserted into the scheduling pipeline without CPU intervention, using notification signals with offset and delta values to coalesce data and schedule processing directly within the processing cluster array.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If CPU intervention is used to insert dynamically generated tasks into the scheduling pipeline, then task insertion can be performed, but latency increases due to communication delays between the multiple processor system and the CPU
Solution Approach 1:
The patent extracts the task insertion function from the CPU and relocates it to the multiple processor system itself. The scheduling unit within the multiple processor system directly receives notification signals from processing units and inserts dynamic tasks into the scheduling pipeline without CPU intervention, thereby eliminating communication delays and reducing latency.
Solution Approach 2:
The multiple processor system becomes self-sufficient by implementing its own task insertion mechanism. Processing units generate notification signals that are directly processed by the scheduling unit within the same system, enabling the system to manage its own task scheduling without external CPU assistance, thus reducing latency while maintaining operational simplicity.
2Speed
If direct communication is implemented between processing units and scheduling unit, then latency is reduced, but system complexity increases
Solution Approach 1:
The patent introduces notification signals as intermediaries that carry task information from processing units to the scheduling unit. These standardized signals include offset and delta values that convey essential task data, enabling direct communication without requiring complex point-to-point interfaces between each processing unit and the scheduling unit, thus achieving high speed while controlling complexity.
Solution Approach 2:
The communication protocol is simplified by changing the parameters transmitted between processing units and the scheduling unit to standardized formats. Notification signals use fixed parameters such as offset and delta values, which reduces the complexity of the communication architecture while enabling fast direct communication for task scheduling.
Data Source
AI summary
One embodiment of the present invention sets forth a technique for enabling the insertion of generated tasks into a scheduling pipeline of a multiple processor system allows a compute task that is being executed to dynamically generate a dynamic task and notify a scheduling unit of the multiple processor system without intervention by a CPU. A reflected notification signal is generated in response to a write request when data for the dynamic task is written to a queue. Additional reflected notification signals are generated for other events that occur during execution of a compute task, e.g., to invalidate cache entries storing data for the compute task and to enable scheduling of another compute task.