A Fine-grained Fault Recovery Strategy for Task Pipeline

With the development of space technology, data processing and computational grow exponentially, leading to future spaceflight hardware platform structure changed from the traditional monolithic processor to multicore platforms. Thus the high-performance computing in space should be considered. Flowing parallel as an effective way to achieve high-performance computing is widely used in the field of space image processing. However, due to the complexity of the space environment, computing tasks easier fails due to failure during running. The traditional recovery policy of simply rebooting the entire system costs excessive computation time although it ensures the reliability of the calculation. This paper presents a fine-grained recovery strategy for the task pipeline model and it ensures the security of the tasks in pipeline with less performance cost. Keywords-failure ; recovery ;task pipeline ; fine-grained