Slashdot Mirror
Carmack on D3 on Linux, and 3D Cards
drfalken writes "Despite recent reports from id management indicating that releasing for Linux is more trouble than it's worth, John Carmack has reaffirmed his commitment in his recent .plan: "It is still our intention to have a simultaneous release of the next product on Windows, MacOS-X, and Linux." This is his first .plan update for months and coming hot on the heels of his MacWorld Tokyo appearance." Actually this part is secondary after his
discussion of 3D cards. Interesting stuff.
I wonder if the economics of the situation could be improved by at least packaging the Win and Linux versions together, and maybe the Mac version too? I can't imagine the number of CDs inside the box makes that much difference to the shelf price.
-Ciaran
I think Matrox is a sitting duck waiting to be hit soon. 2D rawpixel performance is what 3dfx was good at, and nVidia is keeping that brandname and technology alive as well, so I suppose they have more plans for 3dfx and it`s strategy. ATI is much stronger than Matrox, because they can rely on the OEM market which is allways tough to dig into once relationships are going. They also allready have a retail history, and know how to build 3d cards for, so they have the best angle to play nVidia. I hope the can put the pressure on.
But I agree on the self-marketting stragety of nVidia, as you pointed out. The recent Xbox deal must sound like a dream to nVidia`s PR management.
With great power comes great electricity bills.
All your Cacodemon are belong to us.
Rader
Nothing personal, but I cannot disagree more.
I run two machines at home - the primary machine being my Mandrake 7.1 box that I do 99% of everything (graphics editing, document writing, file sharing) on. The other machine, the P-800 "big boy", is Windows 98, and serves one purpose and one purpose alone:
It runs games.
Now granted, I'm a very serious gamer, writing walkthroughs, trying out the latest software, etc, etc, etc. And I don't claim to be any Linux guru at all - I just like the stability of the operating system (when editing 500+ 1 MB JPEG files, it comes in handy, even on a system with half the power of the Windows box).
Recently, I was working on a Linux based review, where I wanted to get Quake 3 running on a Voodoo 5500 card, and review the various distributions on how long it took me.
Long story short: never happened. I couldn't even get the game to run. My fault? Probably - I admit that. I'm not experienced enough in Linux to understand what went wrong.
But games are an important test of an operating systems capabilities - both for how "user friendly" the operating system is, and on how "powerful" the same system is. If the "ordinary user" can't install a game on their box without having to mess with xfree86 configuration files - forget it. With a Windows box, you throw in the Voodoo 5500, slap in the CD-ROM for drivers, and you're done. Quake 3 running in a few minutes in glorious 1024x768 death.
And I'd say we need those ordinary users to make Linux a hit on the desktop. Just because people can run games won't sully the server end of Linux. If anything, if suits who pick up Reader Rabbit for Linux for their kids can see how well it runs (especially when the kids can log into their own sessions so the suit's porn collection doesn't get messed with), it makes it that much more likely to get installed into the workplace. All of the neato configuration files can lurk beneath the surface for us "power users" to tweak if we want, but if somebody wants to change their resolution in X-Windows without going through a gigantic hassle, then let them.
My personal dream is to have every machine in the building I work in, every desktop, server, or otherwise, running Linux to I can fix them remotely if I need to, or just rely on a stable operating system. But do to that, I have to win over the "normal" users - and to do that, they need their games.
Just my opinion. I could be wrong.
John "Dark Paladin" Hummel
52 Weeks, 52 Religions with John Hummel
When I originally discussed what features we wanted in the animation system with the animator, I suggested adding controls for parametric facial animation, and he was basically horrified. His response was that he could do a much better job by hand. "This (animation) is what I do.", he said. After seeing the results of what he can accomplish by hand, I tend to agree.
I've looked into the research that's been done on parametric facial animation, and while it's impressive, I haven't seen anything that approaches the quality that an animator can do by hand. Even when the set of expressions it uses are manually created, the expressiveness doesn't compare to the subtlety an animator can put into it.
While the generality of a parametric system would be great for generating massive amounts of facial animation, as well as animation for dynamic content (such as net-based voice communication), if the animator is willing and able to handle to workload, I am more than happy to stick with hand animation. The technical challenge would be quite enjoyable, but in the end, I'll take a limited amount of high quality hand animation over an unlimited quantity of mediocre computer generated animation.
Jim Dosé
id Software
I've got an Athlon with a GeForce2, running RedHat 7.
/etc/X11/XFree86-4.conf (NVidia is very clear about how to do this... it is slightly annoying) though.
Steps to play Quake3 from box with blank hard drive.
Install Red hat 7.
Install XFree86 drivers from NVidia's site:
rpm -ivh foo.rpm bar.rpm
Modify a specific line of
Install Quake3.
Play Quake3 and enjoy 90+ FPS just like the windows folks do (if you don't believe it, look at how good NVidia's linux drivers are at http://www.tomshardware.com)
This is easier than reformatting and reinstalling everything on a fresh Windows ME box when Microsoft MechWarrior 4 blows up slightly after startup because it didn't like something about the DirectX 8 that was installed. I wish all games ran on Linux, it's easier to learn to use Linux than spend hours reformatting Windows every time something goes wrong.
---
Seriously, I'm a game developer myself (more than 7 years experience of game programming and project management) and after having looked at SDL I find that it contains nearly everything that anyone would need in terms of high performing hardware abstraction for a cross platform tripple A title.
There is also a promising alternative in ClanLib (www.clanlib.org), but I haven't tried that myself.
The people that refuse to pay for Linux software are very often those that refuse to pay for Windows software too. Can you say #warez?
--
Delphis
OK, I'll bite... /. who are getting paid to produce Free Software; and that's only to name a few.
Apache can be produced under the open-source model, things like games cannot - you cannot get graphic and sound artists to work for free...
First of all, who said you had to work for free for Free Software? I know a lot of people at Red Hat, Mandrake, Suse, contrib.net, and even
Secondly, to answer your comment on games in particular, let's see, GLTron is a good, light-weight one to start with; it's hours of pure joy wrapped in a 1.2Mb GPL'd package. Getting into the heavier projects, Parsec (Free as in Beer) looks extremely promissing and will be included in future RH releases, among others.
There's also the much-anticipated Crystal Space 3D engine and SDKs, almost to a first release, and looking incredible; I've been looking to port TacOps to it for a Free Software alternative to my favorite Unreal Tounament MOD.
That's just a couple examples off the top of my head, a quick look at http://sourceforge.net/foundry/games/ would probably be very informative for you!
"I've seen plays that were more exciting than this.
Honest to god... Plays!" Homer Simpson
One of the reasons why Linux is so hard to support is that every distribution of Linux has a different set of standards. They have different directories to put config files in, and different formats for the config files. Distros come with different versions of libraries, some which break binary compatibility between versions. Different shells and window managers make it difficult to help a user install or configure something.
Linux distros need to start following some set of standards, mainly configuration standards that will allow people to help new users better at installing libraries and configuring the system. Until then, Linux is just fighting with itself and preventing commercial software from coming in.
Outdoor digital photography, mostly in New Engl
I'd love to see more of my fave games released for Linux, esp. Half Life.
The Linux Half-Life HOWTO will show you how to play Half-Life flawlessly using WINE. Now, if only I could get my graphics libraries in order under Debian, perhaps I could actually get their WINE package to work and actually play Half-Life, which I bought on Saturday!
Does my bum look big in this?
Part of the problem is where the companies focus their attention. Of the remaining Big Three (ATI, nVidia, and Matrox), nVidia is the only one who actively markets their chipsets to J.Random Cardmaker. 3Dfx started out in the same business as nVidia, then made the fatal mistake of buying a card maker (STB), and trying to leverage the manufacturing plant and distribution channel into selling 3Dfx-branded cards instead of letting others do the work for them.
It must have seemed like a compelling argument to the 3Dfx management at the time - why sell chipsets for around $30 to a manufacturer who then gobbles up the real profit, when we can control all the money ourselves? What they failed to anticipate was that a few things would all converge to crush the life out of 3Dfx as a result:
1- nVidia would dramatically ramp up their product line and take over the performance title.
2- People would start clamoring for 32-bit video before 3Dfx could deliver.
3- Competitive pressures would knock the bottom out of the low-end and midrange video card markets - no profit margin left. And 3Dfx didn't make it into that space in time.
4- Brand-building gets expensive - nVidia doesn't have to advertise at all, really. The card makers who buy nVidia chips do it for them. ATI doesn't have to advertise as much - the retail product is only a piece of the business, they do a ton of OEM. And Matrox has the 2D performance ("business") market locked up pretty well - they aren't playing too heavily in the 3D world.
So with all that figured in, 3Dfx, when the dust settled, had:
A shrinking retail presence.
No significant OEM business.
And no 3rd party card manufacturers to consume the chips.
Each of the remaining players plays to a different niche now - that's why there's still room for three big players. ATI has their All-in-one retail products and the OEM channel, Matrox has 2-D and dual-head for the mainstream Windows market, and nVidia has all the Taiwanese board makers and the gamers market.
The nice thing is that all these companies have to stay on their toes, because one of their competitors could invade at any time. Remember, ATI took a pretty big bite out of everyone when Radeon came out, only to have nVidia make up for it with GeForce 2. Matrox is always rumored to be coming up with super-fast stuff in the labs. So anyone could dethrone nVidia in the future - it's just right now nVidia's sitting fat and happy. They just can't afford to get too comfy up there.
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
I think there was never an issue with making a linux version. They just stated that the economics of putting a linux version box on the shelf are not that good.
We don't have any technology specifically directed towards character features. The animation was done pretty conventionally in Maya.
Our new animator comes from a film background, and we are finding that the skills are directly relevent in the new engine.
John Carmack
What worries me (and Carmack it seems) is the lack of competiton. The only viable competition to nVidia right now is ATI. I really like Matrox as a company (especially their policy on drivers and open source), and have been quite lukewarm towards nVidia due to the driver situation. (Now that they have purchased 3Dfx, maybe they will open the drivers?)
Unfortunatly, it looks like nVidia has a fast product cycle, they release a new card every 6 months. I fear that in a year there will only be one graphics card company. Sure, Matrox and ATI will still be around in the OEM market, but the 3D world will belong to nVidia. That is not good for Linux users, and worse for BSD users.
Here is hoping someone else has something hiding in the wings.
Try to hack my 31337 firewall!
The base for the product is obviously established enough to allow for a rather large penetration on servers alone (check the included server scanner in Half-Life sometime, about half the servers are running on Linux) so why not develop clients?
well, one quite obviously reason would be that the dedicated server requires considerably fewer resources. i have run Quake servers on Linux in the past, but they were all done on an old Pentium 200 at a friend's co-lo.
i have a few boxes that have the hardware (processor speed and video) necessary to run a 3D-game client, but i'm not running Linux on any of them. why? because the 3D graphics drivers on Linux are lousy, X86 is a pain and a bottleneck, and all in all the hardware is completely crippled by the sub-standard driver and API support of Linux. add this to the reasons that Carmack has given (that there's no easy way to port this stuff over to Linux) and you'll realize that it just isn't worth it!
it amazes me how many people on slashdot think that Linux' desktop marketshare is so large: it's not. the majority of Linux users i've met use Linux on their spare equipment, not on their primary desktop machine. it's just not cost-effective to write games for Linux right now. it's a sad fact, but true. sure ID is doing this, but it's really just an experiement. however i believe that the success of D3 on Linux will have a great impact on other game developers' oppions of the platform.
but despite all this, i strongly believe that most of those Half-life servers are on somebody's spare PC, and that the vast majority of those people would never use Linux as their Desktop machine.
- j
I was really curious about something in the video, and I hope John will take the time to answer. It must be a double-edged sword showing that video; on the one hand, it's probably cool to let people see what you're working on. On the other hand, it probably generates a lot of e-mail traffic!
I was really impressed with the mouth movements on the animated faces. I was wondering if you were building in an "expression engine" that could be used on any model, or if those expressions had to be manually animated for each model.
I was thinking that if you specified points on the model face that correspond to the various facial muscles, it would be possible to do a generalized expression engine. For example, specifying the corners of the mouth, cheeks, eyebrows, etc.
I bet that would really save a lot of animating time, and make it extremely flexible to add dramatic expressions.
--
Sometimes it's best to just let stupid people be stupid.
It is said that Linux games are 'more trouble than they are worth'. I think this might be right, but there is a deeper issue here. There is a reasonable-sized market for Mac games, and apparently Linux has a bigger market share than Apple now.
The problem, which hasn't really been discussed properly, is that a minority of Linux users seem unable to recognize the value of software - that some software costs money. The prevalence of free software means that many linux users refuse to pay for software. For many this is an ideological belief - according to the biggest advocates of the GPL, GPL should render software that you pay for obsolete.
This then is the problem for companies considering developing under Linux.
Many Linux users refuse to recognize intellectual property, and so won't pay for games under Linux. This means that the market is fatally undermined - you are attempting to sell to people who won't pay for software, or even believe that paying for software is wrong.
Until people can recognize that software does cost money - that while long-term projects such as Apache can be produced under the open-source model, things like games cannot - you cannot get graphic and sound artists to work for free; then non-free software on Linux will have an unhappy time.
I think this is sad, since there can exist a harmony between the two, but at the moment a minority of unrepresentative bigots, who don't seem to believe that hard work in producing software deserves reward, are holding the two worlds in a deathly battle.
--
Hi!
is put it all on one or two cds. Thinks about it you could just download the tarballs to make Quake 2 work under GNU/Linux they same thing for UT. I seem to remember Carmack saying that Q3 could have done the same thing but they wanted to release a version to be able to track the GNU/Linux adoption. IMHO this was a mistake put it all in one box and make and put a install script on the net (Just like UT has done.) No fuss no muss no overhead added and we GNU/Linux gamers don't have to pay through the nose. This is one of the reasons why I have and enjoy UT and only have the demo for Q3.
Cypherpunks: Civil Liberty Through Complex Mathematics. Those who live by the sword die by the arrow.
A linux release is almost always inevitable in any game that becomes a solid and lasting multiplayer hit. The usual problem is that such a release is often limited to a dedicated server, rarely a client. However it would be foolish not to take advantage of this. The base for the product is obviously established enough to allow for a rather large penetration on servers alone (check the included server scanner in Half-Life sometime, about half the servers are running on Linux) so why not develop clients?
I'll agree that it's nowhere nearly as simple to port a client for a game and attempt retail sales while as it is to release a free server. Who says, however, that it needs to be retail though? I'm willing that bet that there is not a single linux user that does not have sufficient internet access to order the game online, direct from the company if need be. This would allow for a smaller run if nothing else as the product does not need to be shipped out to every retail location.
As to the task of developing the ported client itself? That's where the problem I feel lies. Outsourcing to a company like Loki that deals primarily in ports of game clients to Linux would be helpful, but in-house solutions will be necessary to achieve wide-spread releases of games under Linux.
Maybe the best we can hope to achieve is Mac-like integration, but the installed base willing to run servers and (presumably) buy Linux clients ought to be large enough to warrant more consideration.