The Art of the Tech Demo

Alan writes "A lot of people underestimate the significance of a good technology demo. A good tech demo can be more important for a GPU product launch than even benchmarks. However, this means more than just pretty graphics or complex shaders. In my final article to the industry, I explain what the art of the tech demo is all about. "

Comments

Why don't they just ask the demoscene?

[Anonymous Coward]

We've been doing it for years..

http://www.scene.org

Re:Why don't they just ask the demoscene?

[softwave]

I am not trying to turn this post into a nostalgic troll, I'm actually trying to make a point here.. just sing along with me :)

Back in the days, making a demo was about showing off "software"-skills. A demo was all about coding fancy effects on a broad range of hardware with no support for hardware acceleration whatsoever.
Soft pop, technoish 4-, 8-channel music was tracked (trackered?) on a crappy soundcard (thank you Gravis for bringing us the GUS!!)
Pixelated art was often a result of many hours of hard work, patience and eye for detail.

A tech demo as referred to in the article is much more about showing off the capabilities of new hardware. The demoscene has brought to the fore some very talented artists, some of them making their way into the professional software business.

But still, I think there is a big difference between a demo (as in, "demoscene") and a showoff presentation with some fancy effects. The ideology and philosophy is totally different (as is the expected result and audience).

I'm not trying to be al elitist about the demoscene. I just think that it's a subculture and should be considered as such.

Re:Why don't they just ask the demoscene?

[Archibald Buttle]

Back in the days, making a demo was about showing off "software"-skills. A demo was all about coding fancy effects on a broad range of hardware with no support for hardware acceleration whatsoever.

Back in the days of demos the main demo scene revolved mostly around the Commodore Amiga. It was about showing off software skills, but they most definitely made use of hardware acceleration. The graphics processor in the Amiga ended up doing most of the graphics work in a demo with the 68000 working essentially in a management role. No broad range of hardware on the Amiga though - all the variants were very similar.

Now the PC demo scene inevitably did not have any hardware acceleration as part of their demos, since back then there was none. Video cards in those days on PCs only showed video - even 2D acceleration facilities to speed up drawing windows weren't around... It's only recently with the introduction of more sophisticated GPUs on PCs that some of the kind of things that Amigas could do in hardware are technically possible to do in hardware on PCs. PC demos therefore had to be about software skills rather than what nifty things the hardware could do for you.

Valuable Experience

[eltoyoboyo]

A very successful integrated solution salesman, with whom I once had the pleasure of working, had a very relevant quote for here:

"Do a demo, lose a sale."

The deeper explanation is that so many salespeople come to call with "gadgety" demos and slides. The really successful salesperson LISTENS to a customer's problems and tries to work out a solution in common.

Tech Demos - Dreamcast vs Playstation 2

[JaF893]

The tech demo is how Sony PlayStation 2 was able to stifle the Sega Dreamcast despite platform parity early on.

I think in this case the author is exaggerating the importance of tech demos. I wonder what % of Dreamcast or PlayStation 2 owners have actually seen either of the two tech demos? I think the simple fact that the PS2 was backwardly compatible was much more significant then some stupid tech demo

Re:Misplaced priorities

[doinky]

As a former driver developer, I can tell you that 85% of "driver problems" are hardware problems that the driver failed to work around (sometimes because it was impossible to do so without compromising other goals of the product).

I worked on the Savage 2000 driver for S3, for instance, i.e., the one that everybody thinks was broken because T&L didn't work. Of course, the hardware came back so incredibly shitty that it was actually _slower_ to use its pipeline than the one me and another guy wrote in software (originally as an exercise in load-balancing in high-number-of-lights conditions and to handle a couple of D3D7 features the chip didn't support).

The driver writers at graphics chip companies know their stuff. They're good. Fundamentally, with immature hardware and the desire for speed at all costs, I think they're doing the best they can at this point.