Mirror Magazine

Mirror Magazine Techno Page by Harendra Alwis

How does the processor work?
The microprocessor is an extremely complex device (it is very difficult to explain or understand the details about its inner workings), but its functions can be broken down to simple tasks. The first designs of the modern microprocessor were presented by John Von Neumann; a Hungarian scientist who in 1945 proposed the “stored program concept”; a novel idea at the time. The designs of modern microprocessors have changed very little in principle from the first designs that von Neumann proposed nearly 60 years ago.

All the tasks of the microprocessor can be summarized as performing the instructions given to it by the user and giving an output in return. The microprocessor is analogous to a “little man” (Control Unit) locked inside a room, with only two letterboxes (labelled “In” and “Out”) that let him communicate with the outside world. Inside the room, the “little man” would have a ‘calculator’ (Arithmetic/Logic Unit) and a ‘file cabinet’ with individual slots, each capable of containing a number of limited size (Memory).

It is possible for the “little man” to accept instructions through the “In” box and output the results through the “Out” box and this is essentially what happens in a computer. The “little man” makes use of the memory he has as a temporary storage for the instructions he receives and to store the results before output while making use of the ‘calculator’ to perform all necessary calculations. The “little man” can only perform very simple instructions such as addition and subtraction, and any other task with a higher level of complexity has to be broken down to this level of simplicity.

This simple language that the microprocessor can understand is called “Assembly Language”, but even though it may be simple for the computer to understand, programming in assembly language takes a lot of programming effort. So, in order to overcome this problem scientists and mathematicians have developed “high level” computer languages such as C, C++, JAVA, etc., which are easier to use than Assembly language. You are welcome to contribute your ideas views and comments to technopage_lk@yahoo.com

Multiple instructions at one time
Did you know that a simple program written in Assembly language to print “Hello!” on the computer screen, runs into approximately 17 lines of code? The process of executing an instruction by the microprocessor is done in many stages that repeat like a cycle. These include fetching the instructions from memory, decoding them, keeping track of instructions, calculation, memory management and data transfer. It is a common perception that the microprocessor can perform only one task at a time.

This is not entirely wrong, but even though early processors only executed one instruction at a time, their modern day counterparts can work with multiple instructions at the same time. For example, while the computer fetches an instruction (let’s say instruction a1), it can decode another (a2), and perform the calculations of another (a3) and so on.

This technology (which is not as simple as it sounds) is called “Pipelining” and that is literally what it does; it ‘pipelines’ or links many instructions to the computer at different level so that the computer can work on multiple instructions at the same time. Apart from these, modern microprocessor manufacturers employ various other techniques to beef up their products to make them more powerful and efficient.

Some common techniques are increasing the memory inside the CPU so to minimize the time that is wasted during memory transfer (this will be explained in our discussion about data busses and memory), streamlining the memory management process and using predictive logic (NOT predicative) to effectively ‘guess’ what the next instruction will be. It was mentioned that the processor is capable of performing only very simple tasks.

As the need for making faster and more powerful computers increases some scientists believe that it can be best achieved by enabling the computer to process more and more complex tasks or the CISC (Complex Instruction Set Computers) approach.

Yet others believe that the way forward is to enable computers to process simpler instructions at a faster rate or the RISC (Reduced Instruction set Computer) approach. Both these arguments have valid points but it is my personal view that the best option is to employ the positive aspects of both approaches and to have a balanced approach.

I will not engage myself in an argument as to what brand of microprocessors is superior, but it seems that Intel is leading the market at the moment with AMD a close second. Then again, who can underestimate the power of the Motorola chips that power the iMac G4’s? So with that we end our discussion about the microprocessor and next week, I will introduce the motherboard and the system bus to stage.

Tell me how I feel?
What if a computer could begin to understand what you’re feeling? MIT’s Media Labs’ Affective Computing Group is developing a system that will do just that. Physiological sensors attached to your body and tiny cameras that record your facial expressions lets the computer monitor your reactions.

Then an “affective tutor” will adjust a program to react to your emotions. For instance, if you’re confused by a complex part of a video lecture, the tutor could play it back for you or offer an explanation.

Perhaps the machine itself could express emotion. MIT has developed Bruzard, an interactive animated 3-D character designed to look like a small child. It uses facial expressions to react to your questions. In the future, Bruzard could be hooked up to something like a ‘chatterbot’ to create a more human interface.

Microsoft Research has combined many of these ideas into a concept called Flow. Researchers believe computers should be about giving you back your time. Thus Flow, which is still in the research stage, will allow you to sit at your computer and take part in a virtual meeting. Life-like avatars would represent you and your co-workers so it would look much like a traditional meeting, even though everyone might be in different locations.

And the entire conversation would be recorded and converted to searchable text for later use. One of the big challenges is modelling human attention, so that you could pay attention when you wanted to and not when you didn’t have to. These techniques are a long way from the mainstream. But the combination of animation, natural-language processing, voice recognition, and voice synthesis may very well result in user interfaces that seem more natural than anything we have today.

Back to Mirror Magazine