When technology fails, design wins

The Jakarta Post ,  Jakarta   |  Mon, 09/11/2006 8:25 AM  |  Life

Arnawa Widagda, Contributor, Jakarta

With all that technology has given society, it's easy for people to take for granted the amount of work and research put into PCs and other technical marvels they use every day.

Most of us expect faster, more affordable and capable machines every year. It is amazing how the companies and the army of engineers and researchers behind them are able to make good on that promise.

However, in recent years, it has become harder to maintain the rate of growth as described in Moore's Law. That law predicted, largely correctly, that computer brain power would grow exponentially, doubling every few years.

The most obvious case is the Intel Pentium 4. First introduced five years ago, then latest processor from Intel promised to reach new clock heights unimaginable at the time.

The first Pentium 4 was clocked at 1.4 GHz, which is 400 MHz faster than Intel's own Pentium 3. With the use of a smaller fabrication process, clock speed was quickly scaled up to 2.4 GHz and even 3 GHz.

However, the design proved too much of a challenge even for Intel. A slightly revised design and an even smaller fabrication process were not able to push the Pentium 4 to Intel's target of 4 and 5 GHz.

The reason is simple. It's becoming harder to 'squeeze' out as much performance than before. The transistors are so small at this stage that the electric current needed to switch them on and off is ""leaking"" into others.

Leakage leads to heat and with more than 200 million processors in a single die, the Pentium 4 produces lots of it.

This is why Intel scrapped the Pentium 4 for Core Duo and Core 2 Duo processors. The Pentium 4 relies on its clock to deliver performance, but it is clear that it will not be able to reach 4 and 5 GHz without dissipating enough heat to fry an egg in seconds.

Another example of technological progress grinding to a halt is magnetic storage -- hard drives. For the past 50 years, there has been tremendous progress in hard drive capacity and performance.

Twenty-five years ago, when the first PC was introduced, hard drives could store up to 20 megabytes of data. Now, the latest hard drive can store 500 GBs. We've also seen their dimensions go down, from 5.25 inches to 3.5"" and even 2.5"" hard drives used in notebooks.

Like semiconductor manufacturers, hard drive manufacturers work with a similar and an ever-increasing problem -- how to squeeze more and more data into smaller spaces. Eventually, they will hit a brick wall.

Each sector holding data in a hard drive will become so small, that reading and writing magnetic code in them is impossible.

Drives will lose their magnetism because they are too small and too tightly packed together. This effect is known as super paramagnetism.

Since manufacturers and engineers can't break the laws of physics, they have simply worked around them. Using perpendicular designs, they have been able to squeeze more capacity out of the same space as traditional linear designs currently used today.

Moving to a new design and fabrication process is not without risks. The leading discrete graphics supplier, NVIDIA, learned that the hard way, with its ill-fated GeForceFX line.

NVIDIA was the current graphics leader at the time, practically unrivaled with the demise of competitor 3dfx and no challenges from ATI.

Traditionally, companies have ""gambled"" on new, cutting-edge designs and fabrication process to give their products a competitive edge.

However, this time it didn't pay off. Yields were low and there were lots of performance problems with early designs, so more revamps were necessary. By the time NVIDIA managed to get a working design out, ATI had already been very successful with its Radeon 9700 line and took a large chunk of the market.

Technological hurdles like these are not unprecedented. Every manufacturer, whatever their field, is acutely aware of the limitations in design and fabrication processes.

That's why research is a must for these companies; it allows them to foresee future hurdles before they arrive. Even then, there is always a possibility of an unforeseen problem.

So, what may be considered the latest-breaking technology today will eventually become obsolete. With increasing density and demand for lower cost, it is not hard to see why the computer industry and its supporting businesses have consolidated so much during the past couple of years.

Manufacturers are not only competing with their competitors, but also with the laws of physics.

The consequences of this is not yet being felt by consumers. The price of PCs and other information appliances have hit an all-time low during the past five years and computers are generally more reliable than ever.

It is also not uncommon to see a household with several PCs and information appliances hooked up together into a network.

Consoles, PVR machines, PDAs and even cellphones can all be linked together with the PC acting as the central server.

Instead of having a single machine doing all of our work, we can now split this work across several machines. Therefore, the main driver of technology today is not performance, but cost.

Technology in general has not failed -- it has merely diversified and converged by communicating through a network.

However, failures in technology do happen and like all things in life, one must learn from them and move on.

Copyright © 2008 The Jakarta Post - PT Bina Media Tenggara. All Rights Reserved.