CGDC '97 Roundtable Report

The roundtables I hosted were intended to get together game programmers, hardware vendors, and tool developers to discuss (honestly) the impact of 3D acceleration on the computer game industry. Each roundtable was prefaced by a brief description of my background, followed by a list of the things that I would like to see discussed. This list included:

what hardware sucked
what hardware rocked
impressions of the API situation
what hardware vendors want to see supported by software developers
what software developers want to see from hardware
changes in the way we approach game development because of 3D acceleration
the impact of 3D acceleration on game design, implementation, sales, and marketing

My greatest disappointment was the lack of significant participation on the part of the attendees. Most of the attendees of the first two days were "lurkers", mostly individuals who were attending to get the impressions and opinions of everyone else. Unfortunately, when everyone is like this, no one speaks up. There was a very disproportionate number of hardware vendors to software developers on the first two days - roughly a 4:1 ratio. Of all the attendees, only a very small number (less than three) were close to shipping a game that used Direct3D Immediate Mode.

Uniformly it was agreed that the S3 Virge was the most disappointing accelerator, and that the 3Dfx Interactive Voodoo was the most pleasant to work with in terms of features, performance, and programmability.

The third day was the most raucous, partly because of a large contingent of outspoken Microsoft employees, and also because there was a pretty even mix of hardware vendors and game developers. The debate became extremely heated and very, and the high point was Microsoft stating that "Direct3D vs. OpenGL" has less to do with APIs and more to do with driver models. As a matter of fact, one of the Microsoft evangelists flatly stated that he didn't care which API was used, but mostly cared about the support issues that related to driver models (his contention being that the OpenGL driver models are inappropriate for mainstream consumption).

While I attempted to avoid the issue of Direct3D versus OpenGL, discussions had a way of coming back to this topic, which ended up being the most prevalent focal point of discussion at all the roundtables. There seemed to be a far higher number of OpenGL supporters than Direct3D supporters, which was unexpected, and all but one of the hardware vendors that attended stated that they would support OpenGL. There were also far fewer Direct3D programmers in attendance than expected (only one contributed during all three of the sessions).

All in all I was happy with the roundtables and I think that we learned quite a lot. It would have been nice to get a better mix of software and hardware developers, and maybe next year there will be more real experience that can be shared across the board. Also, I hope that the Direct3D vs. OpenGL debate will have reached some sort of equilibrium by CGDC 1998.