Carbon Nanotube says Hello to us

• 

Researchers at Massachusetts Institute of Technology (MIT) and Analog Devices have created a first programmable 16 bit carbon nanotube microprocessor. The transistors are CMOS logic with nanotube as channel instead of bulk Silicon. This processor containing 15,000 transistor could be a turning point in race to go beyond conventional computing using silicon.

Moving beyond silicon is not easy since no other technology has come closer to the reliability and scalability of current manufacturing process. One can get idea about how good we are at making transistors from the fact that we can add billions of transistors on a single chip.

Although the present 15,000 carbon nano-tube transistor won’t be able to compete with its silicon counterpart but its manufacturing was done with the technology that has been working in commercial world. So, that gives the computing field to move towards future beyond silicon.

Unsurprisingly, this chip was tested with ‘Hello world’ program by Professor Max Shulaker, who is the assistant professor in the institute. It will be interesting to see where does nano-tube based CMOS technology takes us.

Self-assembling chip

Moore’s law (which is actually a prediction) has been a driving force to make computing more powerful and reliable with each passing year. We now have number of transistors equal to global human population on one inch by one inch processor. However, this miniaturisation is approaching the limit set by manufacturing technology and physics. This could act as a bottleneck to recent development in artificial intelligence and big-data.

The good news is that researchers have been searching alternate way to manufacture processors and few like Karl Skjonnemand have got some some success. In particular, Karl’s approach is inspired by nature which has been synthesising complex molecules like DNA with high precision.

Click here to watch TED talk- The Self Assembling Computer by Karl Skjonnemand.

AI program predicts shapes of proteins

Human programmed cell death protein 4, molecular model. This protein is involved in apoptosis (programmed cell death).

Google’s DeepMind project has added another feat in its name. The AI program, AlphaFold has been able to predict the 3D shapes of proteins, which is the fundamental unit of any life.

Researchers have found it hard to understand protein folding and it is not easy to stimulate it in even supercomputers because of high computation demands. Now, AI has the potential to aid researchers and scientist to study proteins in details and get few steps to understand and find a cure for diseases like diabetes, Parkinson’s disease and Alzheimer.

DeepMind trained their AI programme which is a neural network on thousands of proteins until it could predict 3D structure from amino acids. Afterwards, when the program was provided with the input, it used the neural network to predict the distances between pairs of amino acids and angles between chemical bonds that connect them.

This new ability of AI will open doors to new research opportunities in medical science and help us understand the fundamental unit of life which has been an arcane subject for long time.

AlphaFold: Using AI for scientific discovery

AI Accelerator

When we think about the processor, we think about sophisticated circuitry inside our versatile computers which can do a complex calculation, help us make logos, create documents and presentation, play games to name a few.  Designers have thought of processor as a versatile circuit and intended to add more computing power in a single die. However, with new applications like machine learning and artificial intelligence,

Programming language for living cells developed by MIT

Researchers in MIT have developed a programming language exclusively for organic cells to design biological circuits, similar to hardware description language for computer hardware. One can now program the cells to respond to a  particular environment conditions.

The programmers need not have details about genetic engineering and about delicate interaction between cell body and environment. It will allow researchers to develop

Biological Computing

The search for new computing paradigm is encouraging researchers to explore uncharted territories. Researchers have been working on using network-based computing and highly efficient molecular motors. The main advantage of this approach is extreme low power and it could solve the problem with high parallelism.

One could argue that quantum computer also promise of similar promises regarding performance and power consumption. However, we need to understand that quantum computers need almost zero kelvin temperature and on the other hand, bio computing do not fave this stringent demands of operation environment.

In this bio-computer, protein or any other bio-molecular  solve the problem by moving through nano network of channels which is designed according to the algorithm, it intends to implement.

”The biological computing units can multiply themselves to adapt to the difficulty of the mathematical problem”, explains Till Korten, Ph.D. from TU Dresden, co-coordinator of the project. These biological computing units act like processor and memory as they traverse the network.

One can click here to know more about the development.

 

Multi-Chip module GPU

GPUs has played and will play an important role in the development of high-performance computing

No doubt, Graphic Processor Units will be key to develop artificial intelligence and is required in conditions requiring extensive computing resources. GPUs are becoming more powerful with each year, and Moore’s law can be attributed for such improvements over the year. However, with the law slowing down, the number of transistors that can be fit in a die-size won’t grow, leading to a halt in performance improvement. This could impede the progress of many promising technological advancements.

Researchers have proposed Multi-Chip Module GPU (MCM-GPU) architecture based on aggregating GPU modules rather than using single high-performance chip used at

Computational dreaming

In any computing device, we provide the input and the device process the data and provide the data. It works as long there is input provided and won’t work even for a single minute when we stop providing. Consider a system that can work continuously to solve a problem, even when there is no input. It can work on user input when it is available and when the device is not being used, it can work on how to optimise the solutions it provided earlier. It is similar to what humans do. We tend to solve problems at hand, and often we try to optimise it in our mind even we are idle or asleep.

Researchers at Virginia Tech have developed computation method- called computation dreaming, in which device work on real-time inputs during ‘Day phase’ and try to optimise the solution during ‘Dream phase when it no longers accepting any input.

Such computing method can be used for AI applications which require massive

Microchip inside our body

It might seem like a distant future when we would have microchip implanted inside our body for applications like tracking or collecting medical data to identify malignant tumor at early stage. However, that future is closer than we thought. A US based company, Three Square Market has started inserting microchip inside human body to explore what effect it could have in our lives. This radio frequency identification (RFID) chip allows the person to unlock computers and make payment without any card. It is sold by Swedish firm- Biohax international and it could open new application especially in the

Human intestine on chip

Pharmaceutical companies spend billions of dollars to create a drug and considerable amount of time to bring it to the market from their state-of-the-art laboratory. The drugs are first tested on animals like mice before moving on to humans. This arduous journey of drug development has made researchers to innovate to mimic the organs on chip, which could facilitate drug testing directly on human cells without invading actual