Regarding the D3D vs OpenGL statements this week...


The statement itself wan't much of anything newsworthy. His comments were mostly a summary of historical fact, there was no grand reveal. D3D handles several new hardware features and has been giving new features at the cost of backwards compatibility, OpenGL has introduced far fewer innovations over the last decade. This is not news, just summary.

He stated that even though he thought D3D was better in terms of modern functionality he would not be making the effort to switch his code.

He called making the change "a dubious win".



To me that sounds like only a tiny bit of praise for D3D, not a grand statement in its favor.

I think part of the problem is that people put him too high on the pedestal;

Let's give credit where credit is due. Today, many developers learned graphics and other techniques. But back in the 90s, he did push the envelope, effectively singlehandedly (with the team). The original Doom and Quake engines were a complete shift in perspective and made things possible that many dismissed from the start. There was also no internet back then in the sense there is today. One could not just Google for something, be it algebra, API, something math-like. One could not just email the author of an algorithm or an API or OS. BSP and ray-casting, as obvious as they are today, were a breakthrough that made certain things possible several orders of magnitude cheaper. Neither of them were invented for that purpose, but being applied took that extra insight.

Moore's law made most of these obsolete, ubiquity of communication changed the development process.

Technology is specific application of science. As such it doesn't matter how advanced it is, adoption wins out each and every time. Historically, no technically superior product won. Only those that traded advances for simplicity and adoption.

Back around Q3 era Unreal engine was being developed. During that time, it was inferior as far as cutting edge goes and it also performed worse due to bigger reliance on standards (OGL/DX/whatever was back then). Even though it came out with an amazing software renderer that in some cases outperformed 3D cards, it was the long-term aspect that mattered. Q3 went different way, trying to be smart and squeeze out as much as possible here and now.

Instead, Unreal engine focused on tools. It came with top notch editor that could become part of development pipeline from day one. Q3, due to being too advanced, had higher barrier to entry. Unreal won over the long run.

These days, Unreal is far from advanced by design. Engine aims to scale as transparently as possible, so it supports roughly factor 3 in performance difference. Engines like CryEngine or similar scale much further but require considerably bigger input from developers. They need to plan for differences in scale, from slow, memory limited machine to dual-CPU, quad-GPU monsters. And at least part of them will be left disappointed, either due to low fidelity or due to poor performance.

Correct me if I'm wrong, but imho the days of amazing game graphics engine leaps (like how it went from Wolf3D -> Doom, or Quake -> Quake II) are over. The graphics are realistic enough.

I think today the art is more important for the visual quality of the game than the engine that renders this art. And producing this art is as hard, if not harder (= more expensive), than shooting a commercial movie.

He was involved with leaps even before then. His graphics work enabled quite a lot of growth in the early side-scrollers and in early cart racing games.

Just because that particular bottleneck of rendering has been loosened does not mean the man is irrelevant. He is highly connected both as a skilled software engineer and as a seasoned executive.

I'll read his blurbs, just like I'll read the quotes that come from John Riccitiello, Mike Morhaime, or John Lasseter. It's a sad practice to discount people's ideas just because you think they may be a has-been. Learn from any experiences that people are willing to share. This is especially true from those who are now or have been in positions of influence and authority.

I'll read his blurbs, just like I'll read the quotes that come from John Riccitiello, Mike Morhaime, or John Lasseter. It's a sad practice to discount people's ideas just because you think they may be a has-been. Learn from any experiences that people are willing to share. This is especially true from those who are now or have been in positions of influence and authority.

especially since he isn't a has-been, he's just working on amazing stuff that isn't necessarily a blockbuster AAA game; even though rage is looking pretty awesome in large part due to him. His Rage Iphone app is really impressive, AND HE BUILDS ROCKETS AS A HOBBY. It's not like he's doing nothing or producing a bunch of games that are looking worse and worse compared to other titles.

According to wikipedia (granted probably not totally accurate), Crytek has almost 3 times the employees that iD has and Dice has almost twice as many.

Wake me up when games do realtime raytracing with global illumination and produce the same quality as todays rendered movies

This is tangential at best, but I have a severe allergy to problem statements that prescribe a solution. Will it be good enough if we wake you when games produce the same quality as today's rendered movies?

This is actually why Carmack is still relevant and why Dice and CryTech have become such amazing graphics-tech houses: they think outside the box and create solutions that match the need and the capabilities of the time. They create buzzwords, they don't implement buzzwords.

super inteliigent mathmatician he may be but i wouldnt ask him to work with a team of web developers. His code is a total mess. EPIC wrote clean modular slower maintainable code.

if he didn't spearhead his own company with his enthusiasm he would have been that guy who just hacks shit together which nobody else can understand. 6 months down the line the code breaks because carmack got another job.

he is still awesome though.