100 billion is an incredibly big number, and yet a fully developed human brain can have up to as many as 100 billion neurons in it as part of the extensive neural network that provides the brain with the framework it needs to be the amazing mega processor that it is. Much of the focus with A.I. in computers has been to replicate the function of the human brain as best as possible, and to date that’s happened with varying measures of success along with difficulty in measuring the criteria for that.
The key conduits in a brain’s neural network are synapses, and these are literally the bridges between cells along which bioelectrical impulses provide cognitions, impulses, feelings – pretty much and everything and anything that you might have rooted in mental function. Those impulses have their roots in the different cortex centres of the brain, and in much the same way they are found in the chips of computers and the sort. Up until now the relaying function of these chips lost something of its power and authenticity, but that may be a shortcoming that has a fix possibly arriving soon.
New developments of chips modeled after the brain’s neural network are really making waves based on what they can do for expanding on the capabilities of computing devices. Here at 4GoodHosting it goes without saying that this is a topic of interest for any Canadian web hosting provider or any other type of provider that has inherent interests when it comes to making devices capable of more while keeping them suitably compact and usable.
It’s certainly something that can reach out and be beneficial to everyone, especially when you think about what it’d be like to have computers that are as sharp as what we’re all lucky enough to have between our ears. It may be a reality in the not-too-distant future, and so that’s what we are going to look at this week.
World’s First Electrochemical 3-Terminal Transistor
All of this has to do with a new material that has been developed, an electrochemical 3-terminal transistor manufactured with 2D materials. The key component here is a titanium carbide compound called Mxene that takes classical transistor technology into a whole new stratosphere of transmission possibilities for transistors. The relevance is that it is the first electrochemical 3-terminal transistor manufactured with 2D materials.
This is what allows it to function more in line with how a brain would with maintain signal integrity and allowing it to have all the nuanced complexity that it needs to have. With these new chips the electrochemical random access memory (ECRAM) behaves as a synaptic cell in an artificial network, establishing itself as a 1-stop shop for taking data and then processing / storing it. Computers equipped with chips built this way could rely on components that can have multiple states, and perform in-memory computation in ways that would make current capabilities seem pedestrian at best.
Leading to Even More
The further belief among computing science experts is that the MXenes could be fundamental when it comes to developing neuromorphic computers that are closer in operation to human brains and immensely more energy efficient than today’s traditional computers – in same cases thousands of times more energy efficient to go along with all the more detailed and fine computational abilities. A good number of developers will be familiar with CMOS wafer assemblies where layers of 2D are integrated in silicon, and these new chips will do much the same with the 3-terminal transistors. What this will be is a true hybrid integration with the same back of the line processes.
What can be expected? For starters, these super chips would have write speeds that are upwards of 1000x faster than any other ECRAM that has been built to date. If one was to scale 2D ECRAMs to nano dimension the less than a nanosecond processing rate would make them as fast as the transistors in today’s computer so it’s reasonable to think it can fuse into our current computers using CMOS technology process.
That’s due to the 2D transistor metal materials being entirely compatible with CMOS fab process, and the belief is that within a decade users may be able to fabricate special purpose computer blocks where memory and transistors merge to make them at least 1000x more energy efficient than the best computers we have today. AI and simulation tasks could even have 1 million fold energy efficiency for certain algorithms. These new chips are eventually going to be seen in cloud computing services like web hosting and website builders.
The first commercial products with this kind of mega-powerful chip in it may still be a long way off, but industry experts are saying we might see offerings becoming available before the end of the 2020s.
