With NVIDIA’s Tesla Supercomputer selling for around £4,000 each for an entry level version, the price of personal supercomputers is beyond the reach of the average household. But before dealing with the issue of pricing, first of all, why would anyone want so much computing power in their home?

The easy answer is they want a computer that can navigate the web faster, they want web pages to load faster, to see online video at a higher resolution and to see it play smoothly. Plus, they might want to play RAM-intensive computer games, whose graphics are extraordinarily rich and complex already, but on a high-performance computer would be even more amazing.

But as mentioned, prices are too high. That’s the reason most consumers would give if asked if they want a personal supercomputer. But even if they were prepared to pay the high prices, it’s unlikely that personal computers would be made available in large quantities at your local high street shops.

Apart from obvious security concerns, such as the computers could be used by hackers and malware pushers to make their work even more difficult to defend against, there is also the issue of software: there is simply not enough software being made for personal supercomputers.

The vast majority of software makers are concentrating on the current mass-market machines, which can have up to 4 GHz of processing power, while supercomputers tend to have around 1,000 times the speed. Supercomputers used to be the preserve of universities and large computing companies.

For example, when Sony released the PlayStation 3, it was a machine that was capable of far more than what games makers were used to. It took them a while to become familiar with the machine and develop appropriate games for the platform. The machine could have failed commercially, but Sony had already built up a following with the first and second PlayStation – both in the developer community as well as the market at large. As a result of that momentum, the PlayStation 3 became the success it is today.

The launch of the PlayStation 3 was one of the first occasions when the general public was introduced to the term “flops”, an acronym that stands for “floating point operations”, basically meaning a calculation. So a computer that is capable of 5 gigaflops is basically capable of making 5,000,000,000 calculations per second.

The commonly used method of communicating a computer’s speed is by using hertz, which would be more appropriate in the radio industry, referring, as it does, to radio waves. So a computer is said to be capable of a certain number of gigahertz, but it’s unclear to most of us what it actually means, except the higher the number the better it’s supposed to be.

The speed of supercomputers, on the other hand, tends to be measured in flops, although there is a lot more to be considered when deciding which one is the best.

A supercomputer were introduced in the 1960s and Seymour Cray at Control Data Corporation was one of the pioneers. In the 1980s a large number of smaller competitors entered the market, but many of these disappeared in the mid-1990s during the “supercomputer market crash”.

Today, supercomputers are typically one-of-a-kind custom designs produced by “traditional” companies such as Cray, IBM and Hewlett-Packard, who had purchased many of the 1980s companies to gain their experience. As of July 2009^[update], the IBM Roadrunner, located at Los Alamos National Laboratory, is the fastest supercomputer in the world.

The term supercomputer itself is rather fluid, and today’s supercomputer tends to become tomorrow’s ordinary computer. As mentioned earlier, the Apple G5 was promoted as the world’s first desktop supercomputer. But, by definition, supercomputers are computers that are at the very forefront of computing speed and power and the world’s leading supercomputers certainly do not fit on a desktop.

It’s inevitable that mass market computing machines will become increasingly powerful, and stay within a price range that is affordable for most people. Otherwise they’ll fail to find a market. But it’s also inevitable that corporations, institutions and, increasingly, private enterprises will always have the largest, most powerful computers.

According to the top500.org website, which studies these things, the world’s top 10 supercomputers are:

1. Roadrunner, owned by IBM, and running on the Linux operating system, Roadrunner boasts a processing speed of 12.8 gigaflops

2. Jaguar, owned by Cray, it runs the CNL proprietary operating system; it has a speed of 9.2 gigaflops

3. Jugene, IBM, CNK/SLES operating system, 3.4 gigaflops.

4. Pleiades, NASA, SLES, 12 gigaflops

5. BlueGene/L, CNK/SLES, 2.8 gigaflops

6. Kraken, Cray, CNL, 9.2 gigaflops

7. Blue Gene/P, CNK/SLES, 3.4 gigaflops

8. Ranger, University of Texas, Linux, 9.2 gigaflops

9. Dawn, IBM, CNK/SLES, 3.4 gigaflops

10. Juropa, Forschungszentrum Juelich, Linux, 11.72 gigaflops

Cray, IBM, supercomputers Hide

In the computer business, a year is a very long time indeed. So why has NVIDIA’s Tesla Personal Supercomputer failed to become the PC of choice for more netizens around the world since its launch last year? Unlike the iPhone, for example, it is arguably fading into obscurity and the technology is being assimilated into other areas of the industry by other companies.

One reason is the price: a basic Tesla can cost around £4,000 and a high-end flavour can cost two or three times as much. But price alone cannot be the reason. After all, if you build it, they will pay. Apple built the iPhone, priced it at several times more than even the highest spec mobile phones that were on the market at the time, and still the consumers paid.

Could it be the design of the Tesla that puts people off? It certainly isn’t imaginative. The Tesla Supercomputer looks much like any other ordinary PC. So why would people want something that they will have to do their own word-of-mouth marketing for?

“Oh, I see you’ve got a different computer.”

“You can say that again. It’s different alright. It’s a supercomputer.”

“Yeah? What’s super about it?”

“The performance of course. It’s 1,000 times faster than ordinary PCs.”

“Really? But it looks like an ordinary PC.”

“Yeah, I guess they could’ve put some more effort into the design.”

So on the two colleagues/friends/family members go to the specs of the computer, which, needless to say are all very impressive. While the top-of-the-range computers offered by Apple and other companies can boast quad-core architecture, the Tesla can boast that it can house up to 960 cores. Except… it doesn’t boast – you can tell by the design.

NVIDIA aimed the machine at the science market, which may be a reason for the company to opt for a more conservative approach to design. Many scientists, as well as researchers of other kinds, need access to massive processing power in order to simulate experiments that require gargantuan calculations. Which is why most research takes place at universities, an increasing number of which have built their own supercomputers, often using lots of not-so-super ordinary computers.

Scientists are not strongly associated with computer hardware aesthetics and tend to concentrate on function rather than form, unless it’s part of their remit or unless form is part of the function. Like in the car industry, for example, where the more aerodynamic a car is, the more fuel-efficient and faster it is – and the more aesthetically pleasing it is. NVIDIA would have considered all this, and maybe it knew that at that price, it was unlikely to be a mass-market phenomenon like the iPhone, so why bother too much about the design…?

However, it could be argued that innovative yet user-friendly design is the whole reason why Apple is the owner of such mass-market products such as the iPhone, and before that, the iMac, its sleek laptops, and even its G5 desktop computer, which has an industrial-looking casing and was, when it was launched, promoted by Apple as the world’s first desktop supercomputer.

Apple computers tend to be around three times the price of their competitors’ most similar machines. And yet the company has gone from strength to strength in the last 10 years and has spawned many imitators. Maybe NVIDIA should have taken a leaf out of Apple’s design book. Then maybe the Tesla would shift in larger quantities to early adopters, and the rest of us could have benefited from economies of scale and got it on the cheap later on.

NVIDIA could have captured a large share of the gaming market with its machine. It knows enough about that market as its components tend to be sought after by games machines builders. So why the science market? Maybe it thought the scientists would be prepared to spend that kind of money on a computer. But studies show that the average gamer is in his (or her) mid-thirties. And it’s likely that after a home, a computer would be their most important life purchase, even competing with that sleek, aerodynamic, fuel-efficient car they sit in during the constant traffic jams in the city. Maybe they would’ve been prepared to pay a little extra to be in the fast lane when it comes to internet traffic.

apple, g5, iphone, nvidia, supercomputer, tesla Hide