Fast Multithreaded Computations System Implemented in ZYNQ Architecture
Adam Milik, Michał Walichiewicz · IFAC-PapersOnLine · 2022
The paper brings an idea of fast synchronization of multithreaded computations in CPUs. It is dedicated to the implementation of multi-context PLC control or simulation tasks obtained by a dedicated automatic compiler. The synchronization mechanism is implemented as guarded memory cells that enable fast data exchange and seamless synchronization of mutually bounded processing threads. The idea has been verified using Xilinx PSoC ZYNQ architecture. Proposed mechanisms enhance the scheduling of computation tasks and significantly simplify implementing synchronization. The essential idea is based on a specific memory area with data flow control.