A mapping method between polymorphic real-time tasks and polymorphic computing resources

Abstract

The invention discloses a mapping method for multimode real-time tasks and multimode computing resources. The mapping method comprises the steps of judging modes of upper mode management and task attribute decision services according to respectively periodic operation; updating a global task queue; updating task attributes which are changed in the global task queue; determining task priorities according to deadlines of tasks and task critical levels and performing uniquifying on the task priorities; searching a computing resource carrier set available for the tasks according to a task computing pattern; trying to allocate the tasks to the computing resources with the best task processing performance one by one according to priorities of elements in the computing resource carrier set from high to low, and determining the computing resource carrier attributes of the tasks; adding the tasks in a PPC task queue through a PPC processor according to an EDF scheduling method; inserting the tasks to proper positions of a many-core task queue through a many-core processor according to a scheduling algorithm. The mapping method for multimode real-time tasks and multimode computing resources can be used for effectively improving the operating performance of the system, the success rate of multimode task scheduling and the utilization rate of resources.

Description

technical field [0001] The invention relates to the field of embedded polymorphic real-time computing, in particular to a mapping method between polymorphic real-time tasks and polymorphic computing resources. Background technique [0002] The embedded polymorphic real-time computing platform is oriented to key application fields such as aerospace, and the underlying hardware resources, middle-level virtualization resource management software, and upper-level application software constructed for the ever-changing task models and computing requirements of complex embedded applications have A computing system with dynamic reconfigurable, adaptive and intelligent processing capabilities. The embedded multi-state real-time computing platform is composed of four parts: PPC general processor (also known as main control processor), FPGA dedicated logic processor, many-core processor and RapidIO switching module. Among them, the PPC general-purpose processor, FPGA-specific logic pr...

AI Technical Summary

Problems solved by technology

[0009] At present, there is a relatively lack of task and resource mapping for embedded polymorphic computing. FPGAs and many cores are used to accelerate processing of certain applications, which do not involve dynamic change structures.

At the same time, the existing FPGA partition resource management method adopts fixed partition and one-dimensional layout, and the task success rate and resource utilization rate are low; other scanning algorithms have large time overhead and are not suitable for real-time platforms

Images