PCRAM-based data management method for storage and computation integration
Wang Hongyu, Sheng Liming, Xiaolei Tang, Wenqiang Li, Shen Jum, Jun Tao · 2021
The emergence of new storage media technologies speeds up IoT terminal improvement. In this paper, an IoT terminal’s PCRAM-based storage and computation integrated data management method is proposed. The method not only adopts a storage and computation integrated variable index scheme as cornerstone for hybrid memory management, but also expands the traditional FAT file system to achieve management of the storage and computation integrated variables in a file system. Testing results of our prototype implementation show that the performance of the method improves the performance of IoT terminal significantly. Power consumption can be improved more than 20% by using the hybrid memory management of the proposed method in typical application scenario. Data throughput by using new file data access interfaces (sc_store()/sc_1oad()) within the method are also extremely enhanced: in contrast to traditional file wirte ()/read() interface function, the performance of 50kB (and above) file data write improved 140 times and the performance of file data read is improved more than 6 times.