The magazine of the Melbourne PC User Group A View from the Crow’s Nest Mike Chambers

When last our intrepid author stumbled onto these pages, he looked at the future of digital photography. However, it is hardly possible to consider that topic without looking more broadly at the future of personal computing generally. Applications such as digital photography and digital videography are placing strenuous requirements on the next generation of personal computers.

Meanwhile in the corporate world, evolutions in desktop and mobile computing are challenging long-held IT beliefs about how and where people work. So this is a good time to consider the capabilities we need in our next personal computer. But it is also a good time to think more broadly about the very nature and definition of personal computing in the 21st century. And of course, no view from the crow's nest would be complete without a fanciful dream machine designed to make the product engineers cry.

How Much is Enough?

The downturn in the computer market, which began in 2000, brought low prices and high performance to the masses. Since the early days of personal computing, a classy high-end system cost about US$3000. The features of course improved year to year - "more" and "faster". But the price remained relatively constant. The economic pressure of the weakened tech sector has pushed computer and component manufacturers to all time low prices. US based retail chains frequently offer complete, name brand systems for under $700, while systems that push the edge of opulence can be found for about $1000.

How much computing power we really need depends on how we use that power. It's fair to say that most computer owners have more power than they use. In fact today's desktop systems offer more computing power than was used to run an entire Fortune 100 corporation in 1975. For most of its life, it will remain idle, waking up occasionally to browse a Web site, research a school topic, exchange e-mail and so on. This kind of ordinary use might just as easily be done by a 286-based system, were it not for today's modern, bloated operating systems and applications. For three decades now, microprocessors have been able to think faster than I can type. (That gap, I'm sorry to say, is not narrowing.)

Oddly enough, the bleeding edge of power applications is owned now by PC gaming. Rendering 3D images in a 2D space, moving lifelike animated images and reproducing realistic background scenery require enormous computing power once avail- able only with "supercomputers." If you have a gamer in your household, chances are that the fastest system you can buy today won't be fast enough 18 months from now.

This is an enormously significant data point. While the computing horsepower need for gaming is driving the hardware, this segment has driven some of the most significant advances in computer software as well. People like John Carmack (the technologist behind ID Software) began producing software with graphical capabilities that pushed the very edge of graphics software technology - and doing it for a shareware fee. Twenty years later that's still remarkable.

Now let's pause for a moment to consider what has happened here: In 1998 "corporate systems" were the high-end power systems and the "home" market was somewhat underpowered. It provided a wonderful price-point value model for computer marketing mavens who designed product lines to bilk corporate users for their high performance systems while delivering older, underpowered technology to the consumer segment. My... how things have changed. Today, the vast majority of new systems are sold into the consumer segment. "Power Users," once "information professionals" running complex Excel macros, are now 8 year olds running Quake II while mom's not looking and on systems that dare not cross the three digit price boundary. Perhaps we can understand why computer marketers are taking up basket weaving.

Over time, this might change again. The proliferation of broadband Networking is changing the way digital intellectual property is bought and delivered. Most of us by now have purchased software over the Internet where no physical media or manual ever changed hands: a purely electronic transaction. Today broadband has expanded e-commerce to the audio industry, where music can be bought (or stolen) over the Internet. Already digital subscriber services offer real-time video on demand, delivering movies to digital decoder systems.

The digital content industry promises the possibility that soon we will be able to buy or rent high quality, full-length motion pictures with media-less delivery via our personal computers. Yet we know that even at today's limited broadband download speeds, the speed of the personal computer is often the bottleneck for media downloads.

In addition, the growing popularity of digital photography and digital video cameras is pushing mom & dad's computer use back into the "power user" category. Digital video requires massive amounts of storage, the ability to read it speedily, and the processing power to interpret and display the content unfalteringly. So we may be seeing a convergence on the home front for high performance requirements.

Recent thinking in the corporate desktop world has been that users don't need the blistering performance required of the home gaming systems - that reliability and manageability were the key selling points. I'm surprised to see that many computer industry pundits continue to stick by that view. What they have failed to internalise is the significant extent to which the business use of personal computing has changed. I might not need a corporate computer to play MP3s; however, increasingly I'm using my corporate system to monitor industry video broadcasts, catch industry analyst briefings live, and generally use multimedia for business intelligence rather than simple entertainment.

Moreover, pervasive broadband computing in the corporate sector is creating general demand for real-time collaboration and interactive multimedia presentations. These capabilities enable me to sit at my home office in the U.S. and discuss large-scale server technology with a data centre manager in Sydney, Melbourne, or Auckland. So on the corporate side as well, we are seeing a new shift toward high performance desktop computing.

So let's delve more directly into where component manufacturers are headed in the near term and then turn our attention to the changes in the way we use computing power over the next 5 or 10 years.

Today's Desktop

We can't really talk about today's desktop without clarifying which one.

Much of the equipment that sits on the world's desktop today was defined by the Y2K preparation of the late 1990s. (In fact, that enormous spending on new computers, applications, and code remediation was responsible, in large part, for the tech sector downturn in the new millennium.) An informal survey of PC owners and IT managers reflected that over 68% of installed PCs are 2-4 years old. Despite all the advertising of the latest and greatest, most of us are using yesterday's technology. (This article, in fact is being written on a 450 MHz Pentium III system, circa 1999.)

Windows 98 was the prevailing home OS at that time, while NT4.0 ruled the corporate roost. 128 MB RAM was considered tasteful; 256 MB was posh. 5 GB of disk was common; 10 GB was decadent. Most likely there was no Network card at all (especially at home), or if one was there, 10 Mbit ruled the day. USB was the new kid on the block, but only available to Windows 98; Windows NT couldn't talk to it.

What is available today is of course an altogether different story. Three Gigahertz systems from Intel and AMD are the leading edge. I'll pull my example modern system from this week's ad circular: outpost.com is featuring a Sony VAIO RS100 - a 2 GHz Pentium 4, 512 KB L2 cache, 400 MHz front side bus, DVD-RW/CD-RW drive, 60 GB disk, 256 MB PC2100 DDR RAM, USB 2.0 & Firewire (front & back), 10/100 Ethernet, and a boatload of digital photo, video, and DVD mastering software for US$799.

Table 1 puts this is some perspective, depicting the average desktop in use today, the PC you could step out and buy tonight and finally, the system that you'll be able to buy for Christmas 2005.

Somewhere Over the Rainbow

A 10 GHz processor, a terabyte of disc on a single drive, a gigabit Ethernet interface. By Christmas 2005, that's what a high performance system will look like. And hold on to your hat: without a monitor that system will cost US$500 - $800, probably less.

Driving the madcap race to this blistering pace is the processor competition between venerable Intel and its increasingly accepted challenger, AMD. AMD's success in the value PC market and its respectable price performances have forced name brand players to rethink their processor sourcing strategies. It is also forcing Intel to accelerate the pace on its processor architecture development, bringing to the consumer a wider array of choices and unprecedented levels of performance improvement over time.

Within the next six months, Intel should begin shipping the first of its next family of processors, code-named "Prescott." The Prescott family will start at 3.2 GHz and grow in steps up to a 5.2 GHz processor with an 800 MHz front side bus. Following Prescott, Intel appears poised within 18 months to begin delivering on the "Tejas" family of processors, which will start at Prescott's 5.2 GHz high end and grow in small steps up to a 9.2 GHz processor and a 1 GHz front side bus (1066 MHz). From there, perhaps by Christmas 2005, the "Nehalem" family of chips will cross the 10 GHz barrier and provide a 1.2 GHz front side bus.

What remains to be seen is what will change in end-user computing to absorb the massive growth in processor, IO bandwidth, and storage capacities that manufacturers are driven to build.