IBM Announced A revolutionary Analog AI Processor
A state-of-the-art analog AI system.
IBM has announced a revolutionary analog AI processor that is capable of executing difficult computations for deep neural networks (DNNs) with amazing efficiency and precision. This development, detailed in a new research published in Nature Electronics, represents a major step forward in the quest for high-performance AI computation that also significantly reduces energy consumption.
The performance and energy efficiency of deep neural networks are constrained when they are executed on typical digital computer platforms. The continual data transmission required by these digital systems between memory and processor units hampers computational speed and limits opportunities for energy optimization. IBM Research has used analog AI, which is based on the same principles as those found in biological brains, to take on these problems. Nanoscale resistive memory devices, in particular Phase-change memory (PCM), are used to keep track of synaptic weights in this method.
Read Latest blog from us: AI And Cloud- The Perfect Match
By applying electrical pulses to PCM devices, we may create a continuous range of synaptic weight values. By doing computations directly in memory, this analog approach reduces the need for unnecessary data transfer, leading to improved efficiency. The brand-new CPU has 64 analog in-memory computation cores, making it a state-of-the-art analog AI system.
To smoothly switch between the analog and digital domains, each core has a crossbar array of synaptic unit cells and small analog-to-digital converters. Nonlinear neural activation functions and scaling procedures are also managed by digital processing units inside each core. The chip also offers a worldwide digital processing unit and digital communication paths for interconnectivity. To prove the chip’s efficacy, the study team got an astounding 92.81 percent accuracy on the CIFAR-10 picture dataset, which is very high for analog AI devices.
Its greater compute efficiency over earlier in-memory computing chips was highlighted by its throughput per area, which was measured in Giga-operations per second (GOPS) by area. This groundbreaking chip is a major advancement in artificial intelligence technology because of its energy-efficient architecture and improved performance. The innovative design and amazing capabilities of the analog AI chip pave the way for a future in which AI computing that is both fast and light on energy may be used in a wide variety of contexts. The discovery made by IBM Research is a watershed event that will spur the development of AI-driven technology for years to come.