Research And Implementation Of Program Static Partitioning Method On Heterogeneous Processor

Static diversity of heterogeneous multi-core processors integrated processing core types, the type of processor to support a particular application (such as signal processing, image processing, etc.), real-time scheduling problem, in this case, real-time scheduling has become a multi-verify the mainstream in the field of research directions.At present, Scheduling problems which based on heterogeneous multi-core processor architecture is still in its infancy, heterogeneous multi-core processor-based real-time task allocation algorithm has achieved good results, but the results are still the algorithm for the particular case, and cannot be widely used with all types of tasks set, due to the rapid development of the processor hardware architecture, as well as multi-core heterogeneous processor systems in reality a wide range of application needs, research on multi-resource heterogeneous multi-core processor system task allocation algorithm is desperately needed, with reference to the existing real-time task scheduling algorithm for real-time scheduling and implementation, given a specific set of tasks, the efficient implementation of the set of tasks in heterogeneous multiprocessor systems, this topic is also currently the heterogeneous multiprocessor fields urgent problem to be solved.This paper studies the static division scheduling algorithm on heterogeneous multi-core processors, heterogeneous multi-core processor architecture, and extract the characteristics of the need to schedule the task. Eventually come to a task execution system resources as well as the processor core can provide system resources, and comparison between the two, both mapped to the same Euclidean space through the mapping function. Then weighted Euclidean distance between the two to find the shortest "task-core" matching the task to match each other, and the core of the best adaptation to guide the "task-core" matching pairs behind schedule. Finally, this paper the algorithm to achieve through the LLVM Clang front-end compiler task feature extraction, through the intermediate files "task-core" match temporary OpenCL as a master control program, and ultimately to the implementation of the algorithm time speedup compared with the normal scheduling policy to verify the effectiveness of this scheduling algorithm.