Over this weekend, please join me as I delve into the world of "Quantum Computing" and share my personal thoughts on this rather exotic and fascinating subject....
NOTE: For those who do not have time to read the full article, please have a quick glance at the attached illustration where you will see "Raman Rays" named after the great Indian Physicist Sir C V Raman... The contributions to the arguably most complex field of science viz. Quantum Physics in the first part of 20th century by Indian Scientists like Dr J C Bose after whom the GOD particle or "Higgs-Boson" particle is named and Sir C V Raman who is credited with the discovery of "Raman Effect" and the evolution of the intricate theoretical physics and mathematics behind the same are indeed remarkable...
We have all heard of "Quantum Computing" which is the next big wave in the field of computing and is arguably going to unleash the huge turbo power boost much needed in the current computing world...
With the Moore's law already slowing down in its tracks and Microprocessor pioneers like Intel delaying the next release of their chips as they are battling the engineering challenges while trying to fit more transistors in a square nano meter of space, alternatives to Silicon chips need to be actively explored. While Intel managed to release the current generation of Xeon Chips in 2015 with inter-transistor gap of 14 nm compared to the 22 nm in their earlier generation of chips, the next generation of chips with 10 nm gap would probably be released only towards the end of 2017.
And this challenge of the declining returns from the Moore's law will only be all the more accentuated in the coming years and in my minds eye I see the delta increments in technology falling down drastically by 2025...
And if we are to keep producing faster and faster computing chips together with lowering storage costs that are quintessential to make discernible progress in the newer fields of computer science such as Artificial Intelligence and Robotics as well as IoT which will inturn result in huge value addition to the human race, we desperately need the following: Rapid progress in the fields of Quantum Computing (Disruptive Innovation) and Nano Technology (Radical Innovation).
The quantum computers that are being touted as great innovation today are in reality only at a very low level proof of concept stage and cannot be used to do anything useful other than solve a few preprogrammed arithmetic operations. There are 2 sub-categories that are being pursued in the field of Quantum Computing:
1) "Quantum Annealers" which are more akin to the the Application Specific Integrated Circuits(ASIC) of the yesteryears which are used to implement one particular algorithm only and cannot be reprogrammed for any other general purpose tasks.
2) "Universal Quantum Computers" which are the general purpose computing platforms akin to the Microprocessors of today which can be used to perform any general purpose computing tasks and can be programmed on the fly like their Silicon counterparts from Intel today....
Despite the huge amount of hype and hoopla surrounding the area of quantum computing today and claims of huge wins and achievements in this field , all the working prototypes built today ( which work only in carefully controlled laboratory conditions ) belong to the category of "Quantum Annealers" which are capable of running only a single algorithm. However the point to be noted is that if we choose the one right algorithm say a machine vision algorithm from Google or Natural Language Processing algorithm from Microsoft Skype, then that one specific chosen algorithm will run at an unbelievable speed of "100 million" times more in a Quantum Computer than the speed at which it runs on a conventional top of the line CPU available today from say Intel. In the world of AI and Machine learning where the key contributor to the functionality being developed is the computing speed, a "100 MILLION-X" improvement in the computing speed will mean wonders and realisation of unimaginable functionalities in the world of computing....
The picture shown below is from an MIT Tech Review, April 2016, article on the latest Quantum Computer showcased a few weeks ago at a top American University R&D lab. This Quantum Computer is only a "5 bit" computer whose circuitry can handle only 5 bits of data at a time. The circuitry and the quantum chips operate at a temperature of -272 degrees centigrade in the realm of superconductive speeds which brings to the fore certain special quantum properties which make the superlative high speed computing a realistic possibility. Perhaps one would be tempted to compare this '5-bit' computer which runs a "single algorithm" that solves certain mathematical problems such as Laplace and Fourier transformations with the Intel 4004 Integrated Circuit which is a '4-bit' silicon computer that was released a whopping 45 years ago in 1971. One might also be tempted to compare it with the modern day commonly available 64-bit computers that sit on every desktop and confidently conclude that the latest quantum computing invention is no big deal.
However it is indeed a big deal as the overall computing speed depends primarily on the following factors: 1) Processor Bits and 2) Processor Speeds & 3) Processor Parallelism
It is to be noted that the bits processed has increased only 4 fold from 16 bits in the first Intel 8086 Processor released in 1978 which is still the primary Architectural backbone and acts as a template even for the 64 bit Intel Xeon Chips being produced today....
The primary factor among the three key factors that determine the overall computing speed is the raw processing power of the microprocessor chip.
So the state of the art "Quantum Annealer" chip that was showcased in March 2016 and whose architecture is shown in the MIT Tech Review illustration below may be only a 5-bit special purpose circuit which is capable of running only a single math algorithm and not capable of being re-programmed or being any where close to a general purpose microprocessor such as the 8086 the late 1970s but it makes up for all its short comings by running at a speed that is "100 Million X" times that of a modern state of the art Silicon based CPU.....
To put things in context the human brain which is the most powerful and fastest supercomputer known to us today runs at a speed of 1 Billion X times using the complex and intricate connections of hybrid bio-chemical and electrical circuits aka neurons with their superbly tentacled web of connectors called as synapses...And on top of this raw computing biological circuit speeds is the incredible amount of massive parallelism which is simply unprecedented in the computing world....
Assuming that the Moore's law will continue giving the same returns as it did over the last three decades it will take yet another good three decades from today before the silicon microprocessors catch up with the theoretical speed possible via quantum computing and possibly a good 4 decades from today to catchup with the processing speed of the human brain... However as I argued earlier the Moore's law is showing diminishing returns already and will most likely hit a plateau in about a decade's time making the above conjectures null and void...
If the dreams of computer scientists over the last century of building a computer whose capability equals that of the human brain are to be realised the only way is thru' the advancement of the "Quantum General Computing"....
In my mind's eye I see the "Quantum General Computing" evolving to the stage where it can power Robots with near human like intelligences in about decade or so from now....
As the ancient Chinese proverb goes, "May we all live in interesting times..." 😊
Note: One might wonder as to why I have been only talking about the Intel chips as a baseline while comparing Silicon Transistor vs other technologies. What about the cutting edge chips in mobile phones and tablets which are supposedly the in-thing today? Unfortunately the microprocessors and other computing chips used in the so called hi-tech modern gadgets such as tablets or mobiles are no where near the cutting edge of innovation in the field of chip design. Intel still continues to lead the race in the innovations belonging to the realm of silicon computing chip design...😊
#Musings #Technology #Future #Computing #AI