Symbol execution method and system based on image processor acceleration

The invention provides a symbol execution method and system based on image processor acceleration. The symbol execution method comprises a symbol execution decomposition step of decomposing the symbolexecution into a parallelizable part and a non-parallelizable part; and a task allocation step of allocating the parallelizable part to a GPU for parallel execution, and allocating the non-parallelizable part to the CPU for execution. According to the method and the system, the simple instruction analysis process is executed in parallel by utilizing the characteristics of large kernel number andhigh parallelism of the GPU, so that the path explosion problem is relieved, and the symbol execution is accelerated. 本发明提供了一种基于图像处理器加速的符号执行方法及系统，包括：符号执行分解步骤：将符号执行分解为可并行化部分和非并行化部分；任务分配步骤：将可并行化部分分配给GPU并行化执行，将非可并行化部分分配给CPU执行。本发明利用GPU的内核数量多、并行度高的特性去并行地执行简单的指令解析过程，以缓解路径爆炸问题从而加速符号执行。