Article image

Memristors and analog computing take scientists back to the future

Engineers have developed an analog computing device known as a memristor that promises to redefine the landscape of computing by tackling complex scientific computations far beyond the capabilities of current digital systems.

Digital computing, the backbone of today’s technological advancements, is hitting its limits. The challenges posed by intricate scientific queries, from modeling nanoscale materials to unraveling the mysteries of climate science, demand more than what current digital infrastructures can provide in terms of speed, energy efficiency, and computing power.

Breaking through the digital computing ceiling

Developed by a team from the University of Massachusetts Amherst, along with collaborators from the University of Southern California and TetraMem Inc., the memristor is a device that promises to alleviate these limitations and propel computational capabilities into a new era.

Professor Qiangfei Xia, a key figure behind this innovation, sheds light on the limitations of conventional computing methods.

He explains that the traditional approach of shuttling data between memory and computing units is akin to a “traffic jam” in processing, especially for complex tasks that handle large volumes of data.

Memristor explained: Bridging memory and resistance

The team’s solution? In-memory computing with analog memristor technology — an ingenious method that minimizes data transfers, effectively easing the computational bottleneck.

At the heart of this technology is the memristor itself, a device that marries the functions of memory storage and electrical resistance.

Unlike the transient memory of current transistor-based chips, the memristor retains information even when power is cut off, thanks to its ability to remember its previous state. Moreover, it can be programmed with multiple resistance levels, enhancing information density within a single cell.

This technology truly shines when memristors are organized into crossbar arrays, performing analog computing that leverages physical laws to accelerate matrix operations — critical yet power-intensive computations in neural networks.

Leaping forward by going backwards

By executing computations directly at the device site, this approach dramatically reduces the need for data movement, akin to the clear roads during the pandemic peak, as Professor Xia vividly compares.

“You eliminated traffic because [nearly] everybody worked from home,” he says. “We work simultaneously, but we only send the important data/results out.”

Building on their prior success with low-precision tasks like machine learning, the team has now extended the capabilities of their memristor to tackle high-precision scientific computing.

This leap was made possible through a novel circuit architecture and programming protocol, which allows for the efficient representation of high-precision numbers with reduced circuitry, energy, and latency.

The journey and future of the memristor

The significance of this advancement cannot be overstated. Not only does it enhance the potential for neural network computing, but it also opens up new horizons for high-precision scientific computing.

The team’s memristor has demonstrated its prowess by solving complex equations crucial in fields such as fluid dynamics and magnetohydrodynamics, marking a significant step out of their comfort zone and into the realm of high-precision scientific inquiry.

This breakthrough represents the culmination of over a decade of dedicated research, design, and development by the UMass Amherst team and their collaborators.

Professor Xia’s vision for the future is clear. “Our research in the past decade has made analog memristor a viable technology. It is time to move such a great technology into the semiconductor industry to benefit the broad AI hardware community”, Xia says.

The memristor’s journey from a concept to a technological marvel capable of revolutionizing computing underscores the relentless pursuit of innovation. It heralds a future where computing limitations are a thing of the past, opening up new possibilities for scientific exploration and technological advancement.

More about analog computing and memristors

As discussed above, analog computing operates on the principle of continuous data processing, differing fundamentally from digital computing’s binary approach.

Unlike digital systems that manipulate bits in discrete steps to perform calculations, analog computers process information in a way that mirrors the natural, continuous variations in the physical world.

This method of computing excels in tasks that involve complex mathematical models and simulations, such as weather forecasting and aircraft design, by directly utilizing the analog nature of the inputs to produce outputs with potentially unlimited precision.

The memristor revolution

At the heart of the analog computing renaissance is the memristor, a component that effortlessly blends the functionalities of memory storage and electrical resistance.

First theorized in 1971 by Leon Chua and only realized decades later, memristors have the unique ability to retain their resistance level (and thus, the information) without the need for a continuous power supply.

This trait not only contributes to energy efficiency but also allows for the construction of computing architectures where data processing and storage occur simultaneously, in stark contrast to the separate components used in digital systems.

Bridging gaps with memristor technology

Memristors are revolutionizing the computing world by addressing two critical challenges: speed and energy consumption. By facilitating in-memory computing, memristors minimize the need for data to travel between separate storage and processing units, significantly reducing the energy costs and latency associated with these operations.

This efficiency is particularly advantageous for tasks that require the handling of vast amounts of data, such as machine learning algorithms and complex scientific computations, offering a sustainable path forward in an era increasingly dominated by data-intensive applications.

Implications and future innovations

The integration of memristors into computing systems is not just a theoretical exercise; it is a practical solution with far-reaching implications. By leveraging the analog properties of memristors, researchers and engineers are developing circuits that perform computations more akin to the operations of the human brain.

This bio-inspired approach not only enhances computational power and efficiency but also opens new avenues for the creation of artificial intelligence systems that can learn and adapt in ways previously unimaginable.

In essence, the advent of memristor technology and the revival of analog computing promise to usher in a new era of technological advancements. With their ability to process information more naturally and efficiently, memristors are setting the stage for a future where the boundaries of computation are continually expanding, paving the way for innovations that will transform our interaction with technology and the world around us.

The full study was published in the journal Science.


Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.

Check us out on EarthSnap, a free app brought to you by Eric Ralls and


News coming your way
The biggest news about our planet delivered to you each day