Michio Kaku states that AI is not going to happen for 50 to 100 years because Moore's Law is about
20 years from collapsing:
http://www.youtube.com/watch?v=PW8rgKLPHMg
But, Dr. Kaku, the human brain stuffs intelligence in three pounds.  If we simply duplicate the
functioning of the human mind, we will have AGI.

This video is about 100 times as popular as the Memristor Symposium.

Here is a site which discusses the article in Nature,
The missing memristor found:
http://network.nature.com/people/nwerneck/blog/2008/05/05/memristors-and-the-conductive-capacitors-
in-human-memory-units

If you took a semiconductor memory and replaced it with a crossbar array of memristors, one could
manufacture a static memory device which was ten times the density of a similar chip built with CMOS
transistors.  These devices could be used to fabricate memories which could conceivably hold feature
length films in High Definition and fit in a small USB memory stick.

The major barrier to using these devices for replacing the logic functions is that they 'wear out' after
about 10 million cycles.  Once this problem is conquered, I would suspect that the memristor arrays
will be used to construct large neural networks which should revolutionize the pursuit of artificial
general intelligence.  If no solution is found to this 'wear out', then some form of meminductor should
be found because we know that magnetic structures can be constructed that never wear out.

www.computerworld.com.au/article/386594/hp_advances_next-gen_memristor_memory_technology/

HP advances next-gen 'memristor' memory technology

The company said it now understands what goes on inside the minuscule
circuitry elements

James Niccolai (IDG News Service)

16 May, 2011 10:09

HP scientists have made a small breakthrough in the development of a next-generation memory
technology called memristors, which some see as a potential replacement for today's widely used
flash and DRAM technologies.

In a paper to be published Monday in the journal, Nanotechnology, scientists report that they have
mapped out the basic chemistry and structure of what happens inside a memristor during its electrical
operation.

Previously, although working memristors had been built in the labs, scientists didn't know exactly what
was happening inside the tiny structures. So while HP was already confident it could commercialize
the technology, this discovery will allow it to greatly improve its performance, said Stan Williams, a
senior fellow at HP.

"We were on a path where we would have had something that works reasonably well, but this
improves our confidence and should allow us to improve the devices such that they are significantly
better," he said.

Memristors were first described in 1971 by a professor at the University of California, Berkeley. Prior
to that, scientists knew of only three basic circuit elements -- the resistor, the capacitor and the
inductor. Professor Leon Chua posited that there was a fourth.

Decades later, scientists at HP proved that memristors existed, and further proved that they could be
made to switch back and forth between two or more levels of electrical resistance, which would allow
them to represent the ones and zeros in digital computing.

Scientists knew that switching process was taking place but found it hard to study because
memristors are so tiny. HP's latest breakthrough was to use highly focused X-rays to pinpoint a
channel, just 100 nanometers wide, where the resistance switching takes place. A nanometer is
a ten-millionth of a centimeter.

They then mapped out the chemistry and structure of that channel, and thus gained a better idea
of how memristors operate. The paper was jointly published by HP and UC Santa Barbara.

The type of memory that can be built with memristors, called ReRAM, is nonvolatile, which means
devices can retain their data after the power supply is turned off. That's in contrast to DRAM, where
the stored data is lost when the power is cut.

Williams estimated that HP's memristor technology could be commercially available by the middle of
2013, though "that's not an official promise from HP as a company," he said.

HP has built sample devices in its labs that should enable storage densities of 12G bytes per square
centimeter, Williams said That's using a 15-nanometer production process and a multi-level design,
where four layers of memory cell are stacked on top of one another.

Memristors are one of several memory types being developed as potential replacements for flash
and DRAM. As the memory cells on those chips get smaller with each manufacturing process
generation, they are reaching certain physical limits that make it harder for scientists to keep
squeezing out higher densities.

Flash is likely to reach its scaling limits first, and Williams estimated it has only one or two more
process generations left, or about four years at most. DRAM has "a few more" generations left than
Flash, he said.

But memory manufacturers could keep those process generations in use for longer than the 18
months to two years that is normal for each process generation today, he said.

www.zdnet.co.uk/news/emerging-tech/2011/05/17/hp-figures-out-how-memristors-work-predicts-robot-
use-40092797/
NEWS
Scientists at HP and the University of California, Santa Barbara have discovered the physical and
chemical properties of the memristor, a likely component of future memory and computational circuitry.


Scientists have discovered the physical and chemical properties of the memristor, a likely component of
future memory and computational circuitry.

Using highly-focused X-rays, the researchers gained unprecedented understanding of a device they
already knew how to make and use, but without a clear idea of the underlying mechanism. HP hopes
memristor-based memory chips will run at least 10 times faster than equivalent flash memory chips, while
using a tenth of the power.

In the research, published on Monday in the journal Nanotechnology, the team said that it had new
understanding of how current flow caused heating that changed the molecular structure of the device. It
also said it had discovered memristors to be very close in functioning to the neurons that pass
information around the human brain.

Memristors are simple devices that consist of no more than a thin titanium dioxide film held between two
metal electrodes, and they act within circuitry as resistors. However, memristors have the added quality
of remembering the resistance they had when current last flowed through them, hence the portmanteau
name. Their resistance increases or decreases depending on the direction of the current.

Thermal model
The researchers made their discoveries by non-destructively examining the precise 100nm-wide
conductive channel of oxygen-depleted titanium dioxide — actually, Ti4O7 — at the heart of the
resistance switching zone, then feeding that image into a thermal model of the memristor.

The material around the channel, ordinary titanium dioxide, changed phase and thus the nature of the
inter-molecular gap between it and the conductive channel. That gap sets the overall resistance of the
device through quantum tunnelling; the gap's structure was the target of the X-ray investigation.

"One of the biggest hurdles in using these devices is understanding how they work: the microscopic
picture for how they undergo such tremendous and reversible change in resistance," said John Paul
Strachan of the nanoElectronics Research Group at HP Labs in a statement. "We now have a direct
picture for the thermal profile that is highly localised around this channel during electrical operation, and
is likely to play a large role in accelerating the physics driving the memristive behaviour."

According to the researchers, memristors mimic biological synapses and could therefore be used to
reproduce complex networks of neurons, ultimately helping in the development of semi-autonomous
robots.

It is little more than a year since HP Labs announced that memristors could be used not just for memory,
but also for performing logic operations. The company first demonstrated memristors in 2008, but they
were initially predicted by academic Leon Chua in 1971.

In August 2010, HP said it was preparing to commercialise memristor technology, along with the Korean
manufacturer Hynix. At the time, the companies predicted a launch around 2013, saying that memristors
would ultimately be able to replace flash, DRAM and even hard drives.

The problem with keeping DRAM technology going is that it takes up a whole quadrant of the mother
board in a PC and the ReRAM would simply fit under the CPU chip.  DRAM will be totally obsolete in
four years except for legacy systems.

Next