Heterogeneous hybrid memory data processing method and system based on DRAM and NVM and storage medium

• Main Authors: QIN XIAOLIN, GU YONGXIANG, PENG YUNCONG, HUANG DONG, SHAN JINGYANG, CUI XIAOLI
• Format: Patent
• Language: Chinese; English
• Published: 31.12.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a heterogeneous hybrid memory data processing method and system based on a DRAM and an NVM and a storage medium. The method comprises the following steps: respectively configuring the NVM and the DRAM as a main memory and an auxiliary memory; then, utilizing the bidirectional LSTM neural network model with the attention mechanism to perform classification prediction on the data to be stored in the memory to obtain a corresponding prediction result; then, performing memory scheduling control according to the prediction result of the data to be stored in the memory, wherein if the prediction result indicates that the data are fixed data of high-probability read-only operation, the data are stored in the NVM, and if the prediction result indicates that the data are flow data of high-probability write-in operation, the data are stored in the DRAM. Meanwhile, the neural network model is trained by adopting normally read data and normally written data, so that a computer system can more effi