ReRAM Memory With Energy Storage Capabilities
An Israeli company and a French research institute have made an innovative breakthrough in resistive RAM technology, adding the ability to store energy. His proposal would improve power management for in-memory computing operations, using ReRAM memory alongside the main memory blocks.
In-memory computing offers excellent prospects for applications related to artificial intelligence networks and distributed learning, among others. It is based on transferring certain calculation operations to the memory itself, reducing the processor’s workload, and providing processing capabilities to the different network devices.
Different paths are being explored in the development of in-memory computing, and one of them is the use of memories such as resistive RAM (ReRAM), which offers exciting possibilities. one of the Israeli company, in collaboration with French research institute , has further expanded its capabilities, adding the ability to store energy to power in-memory processing.
Their invention combines functionalities such as working or storage memory and energy storage, which they have called “energy in memory”. The idea is that the memory performs in-memory computing operations and stores energy to power these tasks, reducing the overall consumption of the system. And in their announcement, they explain that ReRAM-based batteries are highly scalable and dynamically assignable. Also, they could be placed next to memory blocks, near the processor.
In the opinion of experts, locating the power supply closer to the processor would improve its performance in operations that require maximum power, which generally comes from an external source. And he explains that ReRAM has the potential to act as an energy repository because these devices employ “faradaic processes” to store information within an active volume. Planet says that “it is a unique feature of ReRAM because ReRAM is based on electrochemical processes, surpassing the capabilities of conventional electrostatic capacitors, to the point that he considers this memory as a supercapacitor.
One of the approaches for this technology would be in networks of IoT devices, where a memory node could include ReRAM for in-memory computing calculations and would act as a power store when idle. Planet claims that it is possible to “dynamically allocate memory to be in power storage mode or memory storage mode, depending on what you need to do.” This concept has been demonstrated in laboratory tests, but researchers still have a lot of work to achieve a commercially viable device.