In-memory computing with resistive switching devices

In-memory computing with resistive switching devices

| Daniele Ielmini and H.-S. Philip Wong
The article reviews the development of in-memory computing using resistive switching devices, which aim to address the limitations of modern von Neumann architecture by performing computations directly within memory. This approach eliminates the energy-intensive and time-consuming data transfer between the processing unit and memory, known as the "memory wall." The review covers digital, analog, and stochastic computing schemes, examining the device physics, processing algorithms, and circuit architectures. Key technologies discussed include resistive switching RAM (RRAM), phase change memory (PCM), magneto-resistive RAM (MRAM), and ferroelectric RAM (FeRAM). The article highlights the advantages of in-memory computing, such as reduced latency and energy consumption, but also discusses the challenges, including device scaling, stability, and variability. The authors emphasize the need for further research to optimize device performance, circuit design, and system management to make in-memory computing a viable mainstream technology.The article reviews the development of in-memory computing using resistive switching devices, which aim to address the limitations of modern von Neumann architecture by performing computations directly within memory. This approach eliminates the energy-intensive and time-consuming data transfer between the processing unit and memory, known as the "memory wall." The review covers digital, analog, and stochastic computing schemes, examining the device physics, processing algorithms, and circuit architectures. Key technologies discussed include resistive switching RAM (RRAM), phase change memory (PCM), magneto-resistive RAM (MRAM), and ferroelectric RAM (FeRAM). The article highlights the advantages of in-memory computing, such as reduced latency and energy consumption, but also discusses the challenges, including device scaling, stability, and variability. The authors emphasize the need for further research to optimize device performance, circuit design, and system management to make in-memory computing a viable mainstream technology.
Reach us at info@study.space