From this week's Tech Watch

Why do people like graphics so much?

Graphics companies over time.

We were sipping wine in Seoul, Korea, when my old friend, colleague, and client Neil Trevett of 3Dlabs asked, "Why do people like to do graphics so much?" He asked that after I had told him that I had identified 30 companies offering graphics processors in one form or another for handhelds (see table below).

Neil and I both have gray hair now, and we started reminiscing about the waves we've seen in the graphics business. The first one was the graphics display terminal, which led to the workstation market. An adjunct to it was the simulation market. Then we had the PC graphics market, and now the handheld graphics market. Each of these markets swelled with suppliers and then consolidated as all markets do, and when plotted it looks like the chart above.

And we will see the normal consolidation of the handheld graphics suppliers as well. (Some companies, BTW, participate in all four markets today.)

But none of this answers the question of why. Why have so much effort, energy, investment, and scholarly research gone into graphics over the past 30 years? (It actually began 40 years ago if you start counting with Ivan Sutherland's work at MIT.)

Is it because of the eye-candy effect of graphics? The desire to simulate reality for entertainment purposes? The quest for art and artistic expression? The need to visualize complex scientific phenomena and/or predict them? Or the demand for better information display? Of course, you all know the answer is, yes. Yes, it's all of those things. And yet . . .

We can't find another single segment of the computer business that has attracted so much talent, and money. Supercomputers, which sell for super prices, don't sell in super quantities. The whole market was $5.6 billion last year and expected to grow to $7.6 billion by 2008.

And PC graphics alone last year was almost that much, coming in at $4.2 billion and growing at 12% (that includes workstation graphics). Handheld graphics will be a $400 million market by 2006, not as big as the PC graphics market and yet with over four times as many suppliers.

But graphics chips are not the whole answer, it's what graphics brings to the party. Chips go into boards, boards into computers, monitors are added, software is needed, and so are peripherals.

Get your quick fix

What makes graphics so attractive (no pun intended) is the instant gratification. If you're a graphics chip or algorithm designer, you can see your results quickly—the feedback is intense and so is the rush when you get it right. You simply can't match the sheer excitement of demonstrating a new graphics system. Bring out a new CPU and it's boring in comparison (so then why do CPU transistors sell for so much more than GPU transistors?).

I remember the first time I saw a ray-traced image. It was Turner Whitted (now at Microsoft research) who did it at Bell Labs and we used to talk about how great images were going to look in the future. That's a future that hasn't caught up with us yet, and we're still not doing real time 60 fps high-resolution ray-tracing, but we're getting close. In the meantime we do tricks to simulate the real world; we bump map, and normal map, and texture map, and reflectance map—all those maps are our little tricks for overcoming the lack of processing power to do real-time ray-tracing and radiosity.

The joy of faking it

And that's one of the other joys of graphics-—actually it's two joys. One is the knowledge that too much is not enough in computer graphics and so we have miles to go before our trip ends, if ever. And the other is the clever ways we figure out how to fake it, and how incredibly good-looking our faked results are.

Think about the demos being given on handhelds now, with those tiny quarter-VGA screens that are barely 6 cm. And yet when a developer shows you his or her antialiasing trick we all go wow, that's cool. We appreciate this stuff because we know how hard it is to do, especially in a transistor- and mem-ory-constrained device like a handheld. I was playing "Half Life 2" the other night, cruising through a canal in the airboat when I got distracted (and subsequently killed) by how good the reflections from the sky and walls were passing over the waves and ripples of the water as I moved through it. Now, of course, that violates the first law of suspension of disbelief, but I have a trained eye and it was a work of art, worth dying for—I couldn't not look at it.

And that's what makes our hearts beat faster and investors reach for their checkbooks—that thrill of the image and the implication of virtual reality (a sadly misused and maligned term, but so apropos).

I'd be interested in hearing your views on this subject. Why do you think graphics is so damn interesting? jon@jonpeddie.com