Parallel computing supercomputer

Parallel computing supercomputer or Human brain which is more powerful ? If you are searching on Google, you are using parallel computing supercomputer.

Brief history

Father of parallel processing Danny Hillis gets the inspiration from brain. Intel first launched 4 bit processor i.e. 4004 then 8 bit 8080, 8086, 80186, 80286, 80386, 80486, Pentium I, II, III, IV and the journey is going on. How to build supercomputer ? A single CPU is always slow compared to array of processors working in parallel. Danny wanted to build a computer which would function the way human brain works.

Brain is having millions of neurons which work in parallel to accomplish a task. Danny took this idea and replaced the neuron with the processor. He wanted to connect millions of processors. His experiment “ Connection machine ” was an attempt to simulate the working of brain. He developed the “connection machine” concept based on working of human brain. Our brain has 180 million neurons which are connected to each other. But each neuron is very simple in it’s calculation. He wanted to build a machine with lot of processors which would be able to communicate with each other.

Working principle

Danny applied principles of neurons in brain to develop parallel processing. He started research on parallel processing in 1985 and now the technology is well established. Now he is interested in solving the real world problems with parallel processing.

When he started the research, Amdahl described that it would not be possible to do the parallel processing efficiently. However Danny was confident as he has the parallel processing working in every human brain. What a genius way of thinking? He started thinking to develop the model similar to human brain. With this came the idea of Connection machine. He wanted to develop a computer using connection machine.

He wanted to have the configuration of connection machine similar to human brain. The brain has 180 million neurons which are connected to each other. So he wanted to have as many simple CPU’s as possible with simple tasks. He wanted to build the connection machine with array of millions of processors . All the processors would be connected to each other and communicate with each other.

Let's simplify

Let me simplify this idea. For example, let us assume, there are n nos. of processors. The processor 1 should be able to communicate with 2, 3, 4,…., n-1, n processors. Processors 2 should be able to communicate with 1, 3, 4, …., n-1, n processors and so on. Let us assume there are 5 processors ( n = 5 ).

Processor 1 -> 2, 3, 4, 5
Processor 2 -> 1, 3, 4, 5
Processor 3 -> 1, 2, 4, 5
Processor 4 -> 1, 2, 3, 5
Processor 5 -> 1, 2, 3, 4

In practice, you require large number or processors. Each processor will do simple task. The collective task of n number of processors will be very important in practical problems. It is the collective task of system that is very important. Large number of processors will do the complex task.

You will wonder what is the use of these n no. of processors working together. This array of the processors will be able to solve complex problem. As data increases, the problem of processing data becomes complex.

I will give you one example of parallel processing. “ Google Search Engine ” uses to give the search results to the user. Google uses array of computer processors. It searches billions of pages and gives you most appropriate results. As it has to search millions of pages in very short time, it uses parallel processing. Due to parallel processing, google is able to give results with in short time. As the amount of data increases, parallel processing techniques are getting important. In the world wide web, millions of pages are getting added, it is really difficult to keep track of these pages and give the most appropriate result for the Search query given by user with in a very short time. However parallel processing made it very easy and fast. The google’s patented optimized algorithm with the help of parallel processing computers made it possible. So keep in mind, whenever you are searching anything on Google, you are using power of parallel processing.

Most of the parallel processing computers are used for scientific research where very sophisticated calculations are required. For example, to predict the weather, for biomedical research etc.

Sidenote : Danny built his first computer using dry cell battery , bulbs, swiches and nails to play tic–tac-toe.

However when you compare brain and computer, you will come to know that how powerful is the brain over computer. That’s why human brain could invent the computer.

Our brain can easily identify a picture or face but to accomplish this task with the computer is a hell of a job. The present day computers are very fast but are not intelligent. The challenge is to make them intelligent and to make them think. That’s where artificial intelligence comes into the picture.

The Thinking machines was the Danny’s company which developed the connection machine. Now the company The Thnking Machines does not exist. The hardware part of it is bought by Sun Microsystems and software part of it is bought by Oracle corporation.

Danny took the principle of that brain use “ massive parallel processing” to dvelop a supercomputer in 1985 with 64000 processors. He named it as a connection machine. The connection machine became the foundation for today’s supercomputer. Danny is working on millennium clock which would last for 10000 years. The millennium clock will be developed using simple “ bronze age ” material like rock and mechanical levers and which won’t need electricity. I am sure, Danny will make it a success.

Subscribe to "COMPTECH" ezine to get the latest news and updates on Windows Vista.

Google Search