Roger's Equations Blog

Roger's Equations

This blog is all about science and technology (with occasional math thrown in for fun). The goal of this blog is to try and pass on the sense of excitement and wonder I feel when I read about these topics. I hope you enjoy the posts.

Previous in Blog: New Horizons begins Flyby of Pluto - Daily Updates   Next in Blog: Pluto Flyby - New Horizons Just Finished Closest Approach
Close
Close
Close
7 comments

Memcomputers

Posted July 13, 2015 9:31 AM by Bayes

Memory and Processing

The human brain both stores and processes information at the same time as opposed to conventional computers which store information in a dedicated location and move it back and forth as required by the processor(s). This back and forth leads to what is known as the "Neumann bottleneck", which is a major limitation for modern computers.

But what if computers could be built in such a way that they combined memory and processing similar to how our brains work? Such a computer would be able to avoid the Neumann bottleneck and would be capable of tackling problems beyond the capability of conventional computers. That's exactly what researchers are trying to do, creating what has been dubbed a "memcomputer".

Here is an interesting article I came across on hacked.com talking about the work:

RESEARCHERS DEVELOP FIRST WORKING MEMPROCESSORS, PLAN FUTURE ULTRA-FAST MEMCOMPUTERS

Researchers from San Diego and Turin have developed the first working "memcomputer" prototype, with brain-inspired "memprocessors" able to simultaneously perform computation and memory operations, which could open the way to ultra-fast future computers. The researchers, headed by Massimiliano Di Ventra, a professor of physics at the University of California, San Diego (UCSD), have published their work in Science Advances. The research paper, titled "Memcomputing NP-complete problems in polynomial time using polynomial resources and collective states," is freely available online.

Ultra-Fast Brain-Inspired Computing

Memcomputing, inspired by neuronal information processing in the brain, is defined as computing and storing information within the same units, interacting memory cells called memprocessors. The theoretical concept of memcomputing has been around for some time, but the researchers claim that theirs is the first working implementation. Traditional computers use separate computing and memory units, and therefore must shuffle a lot of data back and forth, which consumes power and time. The brain doesn't shuffle that much information, which is one of the reasons it's much more energy-efficient than current computers.

"To make a quick comparison: our own brain expends about 20 watts to perform 10^16 operations per second," Di Ventra explained to Popular Mechanics, while a supercomputer would require 10 million times more power to do the same number of operations. "A big chunk of that power is wasted in the back and forth transfer of information between the CPU and the memory." A memprocessor operates like a transistor, but also changes some of its physical properties, such as its electrical resistance, depending on incoming data and instructions. Therefore, a memprocessor can perform computation and data storage simultaneously.

Article Continues Here

Here is an article from 2012 on the future development of memcomputers

Reply

Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.

Comments rated to be "almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, rate them!
Guru

Join Date: Apr 2010
Location: About 4000 miles from the center of the earth (+/-100 mi)
Posts: 9086
Good Answers: 1034
#1

Re: Memcomputers

07/13/2015 12:18 PM

I've played around a bit in the past with neural network "simulators", where the neural network node elements are actually computed in a conventional computer. I always wondered when someone would develop a hardware "neuron".

I'm not sure if that is what this is. The "article continues" doesn't seem to work for me.

Reply Score 1 for Good Answer
Guru

Join Date: Apr 2010
Location: About 4000 miles from the center of the earth (+/-100 mi)
Posts: 9086
Good Answers: 1034
#2

Re: Memcomputers

07/13/2015 1:02 PM

The current thinking is that memory and learning in the brain do not take place in the neurons or brain cells but in the connections between them. An artificial neural network adjusts the weights between nodes (neurons) using some learning algorithm. In an electrical circuit, this can be done with varying resistances. Here is a hardware implementation that was designed at MIT a few years ago where the resistors can change their resistance and "remember" the change:

http://www.technologyreview.com/news/537211/a-better-way-to-build-brain-inspired-chips/

Reply Score 1 for Good Answer
The Engineer
Engineering Fields - Engineering Physics - Physics... United States - Member - NY Popular Science - Genetics - Organic Chemistry... Popular Science - Cosmology - New Member Ingeniería en Español - Nuevo Miembro - New Member

Join Date: Feb 2005
Location: Albany, New York
Posts: 5170
Good Answers: 129
#3
In reply to #1

Re: Memcomputers

07/13/2015 4:09 PM

I as well. I'm excited to see this thing in action. By the way, the link to the article appears to be working now. Please try it again when you get a chance.

Reply Off Topic (Score 5)
Guru

Join Date: Oct 2009
Posts: 1449
Good Answers: 30
#4
In reply to #2

Re: Memcomputers

07/14/2015 6:23 AM

I know quite a bit about neurones and their interconnections (synapses). I know nothing about electronic interpretations, but according to this report the missing memristor has not yet (25 June 2015) been found.

Reply
Guru

Join Date: Apr 2010
Location: About 4000 miles from the center of the earth (+/-100 mi)
Posts: 9086
Good Answers: 1034
#5
In reply to #4

Re: Memcomputers

07/14/2015 8:56 AM

I can't say I fully understand this paper at first reading, but I still don't see why a two terminal device that changes resistance as a function of the current cannot exist. An incandescent lamp is an example of a memristor. As the tungsten evaporates from the surface and the filament gets thinner, the resistance gets higher. Granted it is not a linear relationship, especially at the end.

Reply
Guru

Join Date: Oct 2009
Posts: 1449
Good Answers: 30
#6
In reply to #5

Re: Memcomputers

07/14/2015 9:07 AM

I agree that the resistance of the filament will vary according to duration of use. however, the change is in one direction only, which renders it unsuitable as a memory device.

Reply
Guru

Join Date: Apr 2010
Location: About 4000 miles from the center of the earth (+/-100 mi)
Posts: 9086
Good Answers: 1034
#7
In reply to #6

Re: Memcomputers

07/14/2015 11:01 AM

Agreed. It seem some electrochemical device would work, where current either deposits or removes a layer which is resistive, depending on polarity. Another possibility, more complicated, would be a circuit that varies the stored charge on an insulated MOSFET transistor in response to current flow.

Reply
Reply to Blog Entry 7 comments
Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.

Comments rated to be "almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, rate them!
Copy to Clipboard

Users who posted comments:

Bayes (1); phph001 (2); Rixter (4)

Previous in Blog: New Horizons begins Flyby of Pluto - Daily Updates   Next in Blog: Pluto Flyby - New Horizons Just Finished Closest Approach

Advertisement