Knowm, which holds the distinction of being the first company to ever bring a memristor to market, has released two new variants of its device, as well as raw device data aimed at aiding researchers’ ability to develop and improve memristor models.
The memristor, the long sought-after fourth basic circuit element initially theorised in 1971, not only hold “tremendous potential to advance digital computing, they also provide the unique physical properties needed to directly map learning and inference to physical circuits and create extremely efficient AI,” in the words of Alex Nugent, CEO and co-founder of Knowm. “Our growing memristor portfolio and the available data will help to drive the industry forward, providing researchers with the tools they need to develop for this exciting new era of electronics and computing.”
The two new memristors, as well as the company’s previously announced device, are now available in raw die (unpackaged) form with masks specifically designed for research probe stations, which reduces measurement issues that can be introduced in packaging, such as wire-bond contact resistance.
Knowm’s memristors now come in three variants: Tungsten (W), Tin (Sn) and Chromium (Cr), which refers to the materials introduced in the active layer during fabrication. These devices have been designed for higher operating resistance and lower adaptation thresholds, properties desirable in low-power computing applications. Each device has unique electrical properties, which allows circuit engineers to exploit these variations in their designs.
The company is also selling raw memristor data for researchers and engineers. Currently electrical engineers looking to design new memristor circuits typically use idealised models that deviate substantially from physical reality. Having abundant raw data is the cornerstone of accurate model development, and accurate models are critical to integrated circuit design.
Integrated electronics are expensive in part due to the cost of mask fabrication. If a circuit is built to exploit the properties of memristors, but the mathematical models used in simulations are not accurate, the chip will likely fail, resulting in the loss of millions of dollars. “The potential of memristors is so huge that we are seeing exponential growth in the literature, a sort of gold-rush as engineers race to design new circuits and re-envision old circuits,” added Nugent. “The problem is that in the race to publish, circuit designers are adopting models that do not adequately describe real devices. The only way to really fix this problem is to get the raw data out there and compete to develop and improve the mathematical models.”
For more information visit www.knowm.org
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version