Last February IBM researchers managed to create a graphene transistor (a transistor made of carbon layers one atom thick) able to switch 100 billion times per second (100 GHz). That was pretty good and made this technology (still on the lab’s benches) a promising substitute of the bulkier (relatively speaking) and costlier ones based on gallium arsenide and used to power mobile devices like our cell phones.

Rendering of a graphene transistor (the real thing would be too small to see...)

Now, researchers at the UCLA have created a graphene transistor that is able to switch at 300 GHz, three times faster, and that just half a year after the IBM achievement!

What is amazing is that the Moore’s law keeps its validity, over forty years after Moore made his prediciton and after several scientists have said, over the years, that we have reached a limit. Studies, research and ingenuity keep pushing this limit farther away (we know that the physical limits for the Moore’s law lay somewhere 400 years from now).

What is also most interesting is that every time a new technology is created the tagging line is not just “a faster one” but also “a cheaper one” and this is what makes evolution rolling!

We may expect graphene transistors in our hands probably by the end of this decade but in the meantime present technologies will keep getting better.

Researchers have built tools allowing them to look at single atoms and also to move an atom around. Now they have succeeded in assembling, one atom at a time, a transistor. It consists of 7 atoms, four billionth of a metre long. By comparison the tiniest transistor in the latest chip generation is made with over 1,000 atoms, and it is a cube with the edges of 40 billionth of a metre.

The feat was performed by a team of researchers from the University of Wisconsin at the Centre for Quantum Computer Technology. They have used the scanning tunnelling microscope, a tool existing since the 90ies but so far it hasn’t been possible to use it to make working electronic devices at atomic scale.

According to one of the researchers, prof. Simmons, an Australian:

We are testing the limits of how small an electronic device can be. Australia’s first computer was commissioned in 1949. It took up an entire room and you could hold its components in your hands. Today you can carry a computer around in your hand and many of its components are more than 1000 times smaller than the width of a human hair. Now we have just demonstrated the world’s first electronic device in silicon systematically created on the scale of individual atoms. This is highly significant not just for computer buffs but for all Australians. For 50 years, this process of miniaturization has been fundamental in driving productivity growth across the global economy. We have shown that this process can continue.

http://www.physorg.com/news193896845.html

Of course it will be several years, decaes may be, to turn this achievement into a commercially viable production process but still it indicates that we are far from the physical limits where the Moore’s law will crumble. And those limits are beyond the thresholds where the rules of the game for business change. Actually, we are already there.

The dense electronics of today coupled with its low cost has changed the rules of the game in several areas being the force that has led to the emergence of ecosystems. The low transaction cost in many fields.

It is now 7 years since the human genome was (claimed to be) sequenced. Some applications are already available but not as many as the enthusiastic press predicted at that time.

The problem is the cost in sequencing the genome and the fact that if you want to exploit the knowledge of your genome you need to have YOUR genome sequenced. A general one smply won’t do. And this is where the cost factor hits.

However, nowhere else we have seen this cost coming down so fast. The first human genome sequencing cost 300 million $ (this is actually an estimate. The real cost to get to the first sequencing has been 2.5 billion $ plus another billion to refine the sequencing). Last year the cost was 60,000$, that is 50,000 times less! It would have taken 21 years such a cost decrease according to Moore’s law that we see in electronics.

Now, new technologies are promising a sequencing for about 30$. That would represent a further 2,000 times decrease in cost and it may happen within the next 2 years. The 1,000$ sequencing is already on sight.

The A-C-T-G genome bases appear as colored dots

The promise of a 30$ sequencing has been made few days ago by GnuBio, a startup of Harward University. They are applying a technology based on micro-fluidic commercialised by Raindancetechnologies,

www.raindancetechnologies.com . Basically, the genome is split into millions of parts each being encapsulated into a picolliter droplets that is deposited (much similar to an inkjet printer) onto a chip (look at the black and white image showing the deposition of a picoliter droplet onto a microfluidic chip). One million of these droplets can be deposited and analysed per second. The cost slashing is the result of the dramatic reduction in the chemicals required. These are used by the chip to combine with the DNA in each droplet. A light detector picks up the colour of the light passing through the chip. The colour depends on the base being present and an image recognition software capture it and creates the mapping. This process is repeated 30 times to statistically return a correct sequencing.

The time required for a complete sequencing is still measured in days but the sequencing of a single chromosome is a matter of hours.
http://www.technologyreview.com/biomedicine/25481/page1/

The availability of cheap (and fast) sequencing will radically change the practice of medicine and pharma: we are moving towards a personalised medicine with some nice implication for telecommunications on the way.

The storing of the genome for a population like Italy requires a few PB of data (million of GB) and carries along the usual requirements for privacy and availability.

Personalised medicine requires continuous monitoring of the effect of the drugs being administered, hence the shift from a product based offer to a service based offer and the underlying requirement of a network (wireless) that can provide continuous connectivity to a monitoring system.

The Exascale Projects: technologies and infrastructures for tomorrow

Billion of billions of data are being created every year. That is Exascale. In a talk given at PARC on the Physics of Data, Marissa Mayer, Google VP of Search products and User experience, estimated that 281 EB of data were available on line in 2009 (compare this with the 5 EB that were on line in 2002). http://www.parc.com/event/936/innovation-at-google.html . This number is going to increase faster than the Moore’s law predicted increase of processing power. According to HP CEO, Mark Hurd, more data will be created between now and 2013 than the amount of data created by the humanity since the invention of writing.

The increased ratio is due to several factors: more people connected to the Internet, more people creating content and uploading it to the Web (the average person who uploaded data to the web in 2006, uploaded 15 times as much in 2009). But there are also other important factors: a flood of real time data (web cam, to name but one), the higher resolution of videos thanks to greater processing power and, coming soon, the explosion of sensors and the ever more complex structure of the data they provide.

Regarding sensors Mayer placed cell phones in this breed, noting that they are not just pairing up with people, they have, like people, sight (camera), ears (mike and loudspeaker) and senses (touchscreen and more). In the future they will be equipped with vibration sensors, tilt, rotation, location (many already have), navigation, sound, airflow, light, temperature, bio-detectors, chemical detectors, humidity, pressure.

Once you have this kind of ambient awareness, and you can harvest information from million of cell phones, you can leverage that for retail, defence, traffic, weather, climate, social moods…

Of course one of the question is “Who is You?”, and another one is: “Can we make sense out of this avalanche of data?”.

HP is trying to answer both questions by working of a computing platform that , they say, will have the power to deal in real time with massive, distributed data analyses. Google, on the other hand, sees itself as a provide of Exascale data services.

You may want to take a look at the Exascale organization roadmap, http://www.exascale.org/mediawiki/images/4/42/IESP-roadmap-1.0.pdf . They met in Oxford, UK in the middle of April to discuss those two questions.

The game for the management of these Exascale Informatino world is just beginning and clearly Google is well positioned to play a main role, and so is HP (look at the Census project on sensors), but so are, in my view, some Telecom Operators. In order to be significant players, however, they have to help in shaping the complex regulatory scenario, provide an open field for innovation to third parties and understand that their assets are not twisted pairs but information.

Today’s supercomputers crunch tens of Teraflops and Sequoia, the IBM supercomputer said to be ready by 2012 may hit the 20 Petaflops. These kinds of processing powers are achieved by massive parallel computer chips connected with optical fibre and each containing several cores.

A funny, virtual, image of the future IBM Sequoia Supercomputer

There are also supercomputers made up of hundreds of thousands of PCs, and we call them Data Centres. Google and Amazon are probably the most famous ones.Google has a processing power that is already close to a Petaflop (or maybe already over it).

Now, this talking of optical fibre within a super computer is interesting, it rings a bell. What if the various chips are not in the same box (or rack or room)but rather spread out?

True, light is not that fast: it takes about a nanosecond to cover 30 cm, so getting a signal from a chip in Turin and one in Milan takes something like 0.3 milliseconds but if you are in the business of parallel processing you can assume that each chunk of processing may take place within a single chip so that the interchip communications delay can only marginally affect the overall performance.

If we keep on this sort of reasoning, once Sequoia will be out in 2012, will it really be the most powerful processing power around?

Well, consider all the cell phones an Operator like Telecom Italia has around. By 2012 most of them will have over 1 Gflop capacity (actually many millions of them will be well above that capacity). Do your math and you get a processing power in the order of 40 Petaflops, the double of the capacity of Sequoia.

Just imagine if one would be able to harvest this capacity! Well one might say that such a capacity is being used individually by each phone but again, how many phones at any given time are using their processing power? Even considering 2% of them (it is really overshooting also at peak time)you still get over 39 PFlops.

Even better: consider that a new supercomputer happens every two to three years: cell phones are continuously being replaced (85,000 a day in Italy, over 250,000 a day in the USA) so that your processing power keeps increasing day by day.

The crucial factor limiting its availability is energy consumption. However, we can expect significant progress over the next few years (look at the iPad that sports 10 h of continuous video streaming over WiFi) and additionally it may be the case that only a few thousands cell phones may be needed at any given time and by selecting those with better radio coverage (and therefore less power consumption) and most battery available we can optimise the energy consumption over time.

Now, this is the real cloud in the future: the millions of cell phones and devices embedding processing power and communications. There is a completely new set of problems to be addressed, related to massively distributed computing, viral structures and metadata generation. The Telecom Operators have the most comprehensive labs for experimenting, if they just think about it!

The US Department of Energy and IBM have signed a partnership to build a 20 petaflop machine (that is a computer crunching 20,000,000,000,000,000 instructions per second)by 2011-2012 and to follow up with an Exaflop machine (a 1 followed by 18 zeros), providing the processing power of one billion PC of today.

This machine will be able to process the ExaByte of data that are expected to be generated everyday by the Square Kilometre Array telescope project www.skatelescope.org . The project includes the development of a new form of solid data storage, the “Racetrack Memory”.

Clearly this is a starting project and many obstacles will need to be tackled and solved. Also, it is focussing on highly scientific objective and the solutions are likely to be very expensive. Nevertheless, we have learnt that major scientific endeavour generates a fall out of results applicable to the lay man world. I expect that a project aiming at managing Exabytes of data day in day out will create amazing opportunities for our communities in managing and understanding the data they produce.

In 2008-2009 IBM delivered machines sporting a Petaflop processing power. The short term target is multiplying by 20 that power and the 10 year term multiplying by 1,000. This processing power is awesome, however it is what it takes to process the huge amount of data created by the new telescope (and other physical experiments like the LHC are not joking either). It is definitely too much for our everyday needs but also in this area we will see a tremendous growth in data and we will surely benefit from more processing power, may be available on demand in a “cloud” through pervasive, distributed computing. Just think about the personalised medicine where our genome will be used to create the right drug to cure or prevent a desease. What today would take a few months (decoding the genome, analysing its various genes and loci and creating the right protein) should take only a few hours. Even then, we will still require much less processing power than the one targeted to support the SKA telescope.

Major breakthrough are needed in power consumption. Today it will require a nuclear plant to power such a computer. Data transfer will also be a major challenge. Moving around an Exabyte of data per day is equivalent to all data moved around the globe through all telecommunications networks in the year 2000 (voice included, of course).

Researchers are looking into stream computing, a technique to analyse and sort out data on the fly as they are moved around on networks, storing only those that are needed and discarding the rest. Although storage density keeps increasing there is a need for radically different storage technology when you aim at storing EB.

A promise comes from spintronics memories, being studied by IBM, http://www.almaden.ibm.com/spinaps/research/sd/?racetrack

For more info on these futuristic computers take a look at:

http://www.computerworld.com.au/article/319128/ska_telescope_provide_billion_pcs_worth_processing

One of the hot topic in this last year, and probably for two more years, is the cloud, an architecture making the performance of hundreds of computers (and more) available to anybody. From a user perception the user is connected via Internet to whatever computer processing power he requires.

Someone is going as far as to say that in the future all computation will be in a cloud, with users at the edges able to access via the Internet what they want.

I beg to differ in this vision. In my view we are going towards a “cloud world” but “we” are the cloud. The Internet to me will be there to connect the continuously growing functional processing capability provided by each and all of our devices, that we will continue to buy, relentlessly. Clearly, enterprises may see the cloud as a way to dramatically cut their cost, but also in that area I see more a commoditization of applications (many biz applications) more important than the commoditization of processing (or storage).

Now Intel is announcing the cloud on a chip. A concept chip having 48 cores, running at a power comparable to two bulbs (125 W) and flexible enough to be able to activate just those parts of the chip that are needed at any particular time, thus cutting down even more on power consumption. It is made up by 1,3 billion transistors.

The 48 core chip, a cloud on a single chip

My feeling is that this announcement, as the many that I bet will follow in the next decade, will just bring more processing power to the edges, not to the core.

Another interesting aspect that Intel has been quick to point out is that devices embedding chips of this kind will have enough processing power to interact visually with their environment (and we are part of such environment).  This will be changing many rules of the game, and our perception on what it means to interact with objects.

Take a look at the clip with interviews with Intel researchers explaining the chip and their vision of the future.

http://www.intel.com/pressroom/archive/releases/20091202comp_sm.htm

If you feel that chips are too big you will be pleased to know that some researchers is on your same page and is working on it. At the MIT in the Material Science Lab scientists have developed a new technique to grow nanotubes and replace vertical wiring in chips thus allowing more dense, smaller chips.

Structure of a carbon nanotube

Carbon nanotubes are now being produced for many applications but the construction method does not fit with the one of chips manufacturing.

The MIT scientist found a way to produce nanotubes at the same temperature being used to manufacture chips thus opening the way to grow them inside the chip.

The various components present in a chip are connected one another with tiny copper wires but as the chip dimension shrink so the wires have to shrink. Below certain dimension the copper conductivity suffer and the wires may no longer work. This is were carbon nanotubes can help.

Take a look at the article if you like more details on this new technology
http://www.physorg.com/news171812351.html

Moore’s law appears to be on track to keep going for several more years, as scientists keep finding work around to what seemed just few years ago impassable stumbling blocks.

This means cheaper and more performing chips, more innovation and further dissemination of chips everywhere, smarter objects and more communications. The right mix to generate ecosystems.

 As electronics shrinks, the power applied to the various components decreases. This is good in terms of energy consumption but the signal gets so weak to be almost indistinguishable from the background noise. This is considered by some as the brick wall that will stop the Moore’s law.

It is therefore interesting to note that researchers at the ”Instituto Tecnologico” of Buenos Aires, Argentina, have looked at how Nature copes with noise in signal transmission in living beings to solve the problem.

Apparently, noise can be used to reinforce the signal, rather than swamping it. A phenomenon known as stochastic resonance by physicists is being used by neurons to do just that.

Stochastic resonance to reinforce signal leveraging on noise

The Buenos Aires researchers are proposing to use this technique to improve storage. They demonstrated the concept using two resonators and showing that in effect the signal is reinforced by the ambient noise. Now the problem is to have these resonators working at the nanoscale of the storage cells components.

For more details take a look at:

http://www.technologyreview.com/blog/arxiv/24366/?a=f

What interested me is the fact that again and again as we are confronted with a barrier that is impossible to overcome…some researchers find a way around it. I am pretty confident that we will see the continuous evolution predicted by Moore to hold for the next decade (up to 2015/2016 scientists have the solutions required).

What is interesting, of course, is that already today we are seeing that the increased performances have brought us beyond some thresholds where biz rules change and as the increase in performance (and the related decrease in cost) continues in the next decade we will see many more thresholds being overcome and more and more changes in the rule of the game.

Researchers at the Arizona State University have found a way to create a diode (the bit equivalent in electronics) using a single molecule. The drive towards smaller and smaller elemental component has been on for the last 40 years. Now miniaturization has reached the nanoscale (40nm is now an industrial process with 20nm on sight in the next decade and possible 4nm by the end of the decade). A single molecule is below the nm size but one has to remember that to get the number of molecules needed at a, say, 20nm scale one has to consider that they are arranged in a volume, non on a line. Hence, to create a component based on a 20nm scale with a 0.2nm molecules you need 100³ molecules, that is 1 million molecules. Note that this calculation needs to be taken with a grain of salt since many of those molecules also serve the purpose to create a substrate, still we are talking, even at that incredibly small size of a huge number of molecules.

It comes as a staggering advance, then, the results of these Arizona researchers, led by N.J.Tao, that have shown how to create a diode using just a single molecule. The feat is accomplished by using asymmetrical molecules that respond differently depending on the interaction applied.

Asymmetrical molecule used as a diode

The technique developed by Tao’s group relies on a property known as AC modulation. “Basically, we apply a little periodically varying mechanical perturbation to the molecule. If there’s a molecule bridged across two electrodes, it responds in one way. If there’s no molecule, we can tell.”

It is interesting to note that this group of researchers operates in the Biodesign Institute, that is the department looking at how Nature works.

The application of these discoveries are still far away (10 years?) but it is nice to know that we have still a way to go ahead of us.

http://www.sciencedaily.com/releases/2009/10/091013110042.htm