WAS: Open Source 3D Games

[Justin Dugger]

Sorry for missing the bulk of the conversation I apparently triggered,
but I just got finished with a ten hour work session cleaning computer
labs. The one week we actually do something a semester, and the list
actually brings up something I care about. So, onwards to the reply.

> So is the linux kernel playing catch up with some
> other kernel/os? I thought that Linux was pretty much
> a frontrunner. 

The Linux kernel is a front runner, but the top is crowded.  Linux can
be most likened as an experimental kernel, willing to forsake
stability in many cases, and generally not submitting to a strong and
intensive testing unit (some distros may do such work, however).  The
SPARC port is generally considered faster on the same Sun Hardware
because the Sun kernel does a lot of assinine checks and redundancy. 
At least that's what I got out of the flamewar between the SPARC
maintainer and a Solaris engineer.  Anyways, the linux kernel is one
of those things that I mentioned earlier, that gets 100+ developers
looking at it. If you had that many people working on a regular kernel
endevor, I don't think any management team on the planet could handle
it, without resorting to the same system in place now.  Half of what
makes OSS work is the internet centric systems they utilize.

> Also, I've noticed that certain GUI
> aspects of certain OSS window managers were finding
> their way into other places like the Windows desktop.

As best I can tell, you're referring to virtual desktops and pagers. 
This would be an example the imitation rule being a two way street.
It's easier to copy than to innovate.     Even so, virtual desktops
are a kludge.  I rarely use more than one. I would rather see
something akin to Expose, which uses computing power to manage and
display these things effectively. As a tie-in to the major theme of
the thread, rendering windows to images via GL will require some
advanced 3d hardware, although probably not anything more than you
already own.

> Also, if games need to stop supporting such big and
> powerful cards as Radeon 8500s and 9700s then you
> might want to take a look at what is wrong with the
> game software and not the hardware or OSS drivers.

That is a terribly convient viewpoint, but you might want to examine
where that puts you.  You're against more video ram, which means we
may never see expansive landscapes of high quality without a draw
distance and fog.  I agree that Battlefield 2 is overly aggressive,
and might even suffer from poor coding, but there's still other games
that don't run at a reaonable framerate on today's affordable video
cards.  You've effectively taken a stance against progress.  Also, I'd
love to know why you think Carmack's Doom 3 engine is sloppy code.  If
anything, I'd argue that the games that STARTED this threat might be
the "sloppy code:"

System Requirements
200mb of hard drive space
High Quality - Realtime lights and shadows on, bloom on, high detailed
maps, 1024x768 or higher res
- A 1.5-gigahertz Intel Pentium 4 chip or AMD Athlon 1500
- 9600ati or 5700fx
- 256 mb of ram


Given what their "High Quality" results look like, you might be easily
persuaded to that conclusion (the truth is, the models and level
design are 90 percent of how a level looks; the other 10 percent is
about taking the latest tech and helping the artists manipulate it
appropriately).


> Seriously though, most of the "3rd party" OSS writers
> are professional coders. 
If I were a lawyer, I don't think I'd conceed that point. I'll allow
it, if only so I can make statistics up too.

> I would say they are at least
> as capable of writing drivers as the manufacturers,
> and maybe a bit better, since they don't have to deal
> with a corporate boss telling them it needs to ship
> yesterday. 

And yet the mantra of Open Source code is "release early, release
often." At least the successful ones.  I have yet to figure out a cute
one liner to summarize RMS's software. The best I have so far is "brag
early, release eventually."  But I'm being overly bitter, probably
because I'm up past my bedtime.

I guess I'm misunderstanding something here. If these professional
developers are capable of writing such software, then surely by their
professional nature, they could lend their services to the
manufacturers, much in the same way one dude wound up writing the
Atheros driver I'm currently using
(http://cvs.sourceforge.net/viewcvs.py/madwifi/madwifi/README?view=markup).
 Another guy got a Google Summer of Code project to work on writing an
open source version of that driver's hardware access layer, a binary
only piece of code that limits, in part, which frequencies can be
transmitted on, and how much power can be used to accomplish the task.

Also, when I say few dozen contributers, I means people submitting
patches, not just core developers who actively work on features they
think people want and generally move things forward.  And even the
core contributers may be putting in only a few hours a week towards
the task.  Especially when their real job starts getting in the way.
I'd wager plenty of one or two man driven projects stagnate after a
couple weeks of crunch time at a real job pre-empts personal projects.
I know it happened to gnocatan.  Games in particular seem to suffer
from half-done syndrome.  I guess we gamers are a lazy breed =).  This
is another time where the Open Source system can come in handy, when
someone can easily fork a stagnant project and begin progress anew.


> so an OSS project with a few dozen coders is likely to
> out-perform *any* commercial vendor. 
A few dozen coders is already a step above the average beginning
project. A few dozen core coders brings you into the realm of the top
twenty most active OSS projects.  Browsing CVS commits, Inkscape, a
fairly popular and significant undertaking to create something similar
to Adobe Illustrator, has about five core coders. Judging from the
numbers the Inkscape Project status released
(http://www.inkscape.org/status/status_20050801.php), I'd say that's
about accurate.  In the past two months they've collectively added
about 10 thousand lines of code.  I'm not a graphics artists by
nature, but it sounds like you might be qualified to examine the
differences in quality between Inkscape and Illustrator.

Any theory on why OSS can succeed needs to account for the fact that
the biggest successes are (or started life as) mere imitators.  A
Photoshop clone. A Quake clone. A Microsoft Office clone. A Netscape
clone.  A UNIX clone.  My best guess is that people with truly
innovative ideas don't often feel that the GPL (or similar liscenses)
provides adaquate reward, or that the Open Source volunteers in the
community can   adaquately work to improve something with even more
unheard of ideas (this sort of thing being itself a reward from the
GPL).  But I'm hardly a qualified economist.

> On that note expect the SuSE
> distro to get more and more sophisticated and have
> better and better support for hardware. If I am
> reading the currents right, SuSE is heading in the
> direction to get more and more big industry names to
> port code to SuSE. I may be switching back to SuSE
> someday soon.
The beauty of the GPL it's very difficult for SuSE to maintain some
strong advantage in hardware support.  But I get the feeling you might
be referring to Adobe.  This is where I own up to one of the big
disadvantages of using closed code.  By taking the easy way out with
nvidia drivers and the madwifi halfway closed driver, I'm taking a lot
of the steam out of open source alternatives.  It's a difficult
decision, but the situation appears to be solving the atheros problem
on its own, and I don't think nvidia cards will ever see worthwhile
open source 3d drivers. As far as I know, nobody's even tried to start
on such a project, which is partly why I'm amazed the ati 3d oss
drivers fare so well. I've heard on Slashdot that ati might have
contributed that code, which while it would explain much to me, it's
attribute slashdot comments authority. Is this correct, or is the ati
driver scenario even more amazing than I realize?

Anyways, I've probably written enough to persuade all involved to
simply ignore the thread, when the simple goal of the thread was to
discuss 3d games.  I think everyone here should give crack-attack a
shot, even if all they've got is a TNT2 (it ran fine on mine last
year).  Even if you suffer from motion sickness, this game shouldn't
affect you.

The End

Justin Dugger