Stream-oriented fine-grain asynchronous memory access schedule
面向流体系的细粒度异步访存调度*

This paper focused on the asynchronous schedule mechanism, and presented a fine-grain solution to improve further efficiency. This solution involved semaphore and spinlock, to enable the cooperation of heterogeneous cores and the promotion of local speedup. This solution was experimented on the stream prototype system with a set of testbench, to evaluate the performance of the schedule solution, and analyzed related factors.