Forget digital. Is the future of A.I. analog? It is the claim of Mythic, an A.I. chip company, that it can take a leap forward in performance in power by going back in time. Sort of.
All computers had been around for as long as computers have been around, and the world's first room-sized, electronic, general-purpose digital computer buzzed to life in 1945.
Variable range is used as a way of representing desired values in analog computers. In an analogue computer, the numbers are represented by currents or voltages, instead of the zeroes and ones that are used in a digital computer. The beginning of the end for analog computers was when digital transistors won out, but analog machines stayed around until the 1950s or 1960s.
Tim Vehling, senior vice president of product and business development at Mythic, told Digital Trends that digital kind of replaced analog computing. As a result, analog went away.
The reports of the death of analog computing may have been overstated. The beginning of the end for analog computers may have been represented by the triumph of the digital transistor.
It is not building purposely retro tech. This isn't some steampunk startup operating out of a vintage clock tower headquarters filled withTesla coils; it's a well-funded tech company, based in California and Texas, that's building MythicAnalog Matrix Processors.
The M1076 is a single-chip analog computation device that promises to bring in an age of compute-heavy processing at impressively low power.
There is a lot of interest in making the next great A.I. processor. There is no question about that.
The approach isn't just a marketing trick. Moore's Law, the famous observation made by Intel co-founder Gordon Moore in 1965, claims that every 18 months the number of transistors able to be squeezed onto an integrated circuit doubles. Over the past 60 years, this observation has helped to bring in a period of sustained improvement for computers.
Moore's Law is facing challenges in physics. The physical limitations of trying to shrink components have slowed the advances. One possible way around this is through approaches like optical and quantum computing. The analog approach of Mythic seeks to create compute-in-memory elements that function like tunable resistors, supplying inputs as voltages, and collecting the outputs as currents. The idea is that the company's chips can capably handle the matrix multiplication needed to enable artificial neural networks to function in an innovative new way.
The company explains that they use analog computing for their core neural network matrix operations. There are several advantages to analog computing. It is efficient because the neural network weights are used in place as resistors. There are hundreds of thousands of operations occurring in parallel when we perform one of these.
There is a lot of ways to tackle the problem of A.I. computation. We believe that the Moore's Law approach is not viable anymore. It is starting to prove out. Companies will have to find a different approach to make next- generation products that are high computation, low power.
This problem is going to have a big impact on the advancement of A.I., especially when it is carried out locally on devices. Some of the A.I. we rely on on a daily basis combines on-device processing and the cloud. It's like having an employee who is able to make decisions up to a certain level, but needs to call their boss to ask advice.
This is the model used by smart speakers, which can carry out tasks locally, but then send the actual spoken word queries to the cloud.
Some tasks require instant responses, but that is all well and good. As A.I. gets smarter, we will expect more and more of it. There are a lot of applications that need to be applied immediately.
The A.I. chips need to keep up. The cameras are getting better all the time. The amount of resolution data that models for image recognition need to be able to carry out is increasing.
Add onto this the growing expectations for what people believe should be extractable from an image, whether that is mapping objects in real-time, identifying multiple objects at once, and figuring out the three-dimensional context of a scene.
Whether it's for offering more processing power while keeping devices small, or the privacy demands that require local processing instead of outsourcing, Mythic believes its compact chips have plenty to offer.
We are currently in the early stages of commercialization. We have a number of customers that are evaluating our technology for use in their own products, and hopefully by late this year, early next year, we will start seeing companies utilizing our technology in their products.
He said that it is likely to be in enterprise and industrial applications. Don't expect consumer applications to lag far behind.
We will see consumer tech companies adopt our technology as well.
Isn't it about the most perfect meeting point of steampunk and cyberpunk if analog computing turns out to be the innovation that powers the augmented and virtual reality needed for the metaverse?
The company's name would have us believe that the chips are real.
All Rights Reserved © 2024
