Mesa 5.0 Released

Eugenia writes "Mesa 5.0 has been released. It implements the OpenGL 1.4 specification." There's more information as to what's been fixed/added/changed on their SF.net project page.

Comments

cute

[sanermind]

Cute description in the changelog:

- OpenGL 1.4 support (glGetString(GL_VERSION) returns "1.4")

So that's all it takes, eh?
<grin>

How does it compare on windows?

[Max Romantschuk]

It'd be interresting to know how this release compares to other OpenGL implementations on Windows. Anyone looked into this?

Why Windows? It's always interresting to see how any open software solutions stack up versus their proprietary cousins on a proprietary system.

Re:How does it compare on windows?

[The Original Yama]

I personally would like to see how this compares to the OpenGL implementation in the nVIDIA Linux drivers. Anybody got any benchmark figures?

Re:How does it compare on windows?

[dinivin]

Why would you want to see that comparison? nVidia's Linux drivers are hardware accelerated. Generic Mesa is not.

Dinivin

I don't know what they did

[RomikQ]

But the new mesa seems to have intelligent workload distribution between the cpu and the gpu, i e

glxgears running in a small window - 200 fps, average 2% cpu load(with Mesa 4.1 it was 800 fps 100% load),
running maximized in 1600x1200 - 80 fps, 100% load(exactly as with Mesa 4.1).
And all the games and etc run at exactly the same speeds with less cpu load.

All I can say is this is great - nobody needs insane fps numbers above 100 and it saves cpu for my poor apache running in the background :). Server gaming woohoo!

Re:mesa sucks compared to dx 9

[quigonn]

Hey, OpenGL is at least standardized, something you can build on for years. Nobody can guarantee you that the next version of DirectX will be compatible with the current version.

In fact, there are only two 3D APIs that are standardized and (more or less) widely used: OpenGL and OpenInventor.

in the long run, that will change

[g4dget]

A good game generally needs a good budget to back it.

A flashy game with a lot of graphics may need a big budget. But those games are not necessarily good, nor do good games necessarily need flashy graphics. For example, many of the movie-tie-in 3D games are financed heavily and anywhere from mediocre to horrible. On the other hand, excellent games like chess or go are of utmost graphical simplicity, and they have been refined over centuries and millenia in a process akin to open source. Furthermore, there are quite a few excellent open source games with minimal graphics and excellent gameplay.

Closed source, heavily financed games satisfy a yearning for novelty. They spend a lot of money on eye candy and pushing technology to the limit. But really good game design is a long-term, open process. For computer games, that has barely begun. I suspect that in another few years, you are going to see open source games whose graphics is simpler than Doom but whose gameplay beats anything commercially available. And closed source games won't be able to compete with that because they simply can't have thousands of game players contributing directly to the evolution of the game.

Open source is slow--but eventually, it gets there, and it usually ends up doing a better job.

Re:in the long run, that will change

[g4dget]

Open source development is design-by-committee by its nature.

No, it's implementation-by-contributors, which is very different. Design-by-committee is what happens in many companies: a lot of people who don't have to do any of the real work sit around and talk a lot, then dump some non-sensical specification on their programmers. Design-by-committee is bad. Implementation-by-contributors is good.

It takes a long time and if it ever finishes it's a compromise.

Making a good game or good piece of software always takes a long time--because it requires extensive feedback from users. Open source is actually better at that because the users are the developers.

As for being a compromise, of course it is. However, if it's a compromise a substantial number of people can't live with, the project forks. This, again, is good. Closed source doesn't have that option: just because many people think MS Office sucks doesn't mean they can take it, split it off, and fix it.

who cares about "the lead"?

[g4dget]

i agree. OpenGL lost the lead a long time ago.

So? Who cares? I'd much rather use an open API than some snazzy, proprietary thing.

Wait for 2.0 to come out. MS is going to lock Linux out of 2.0 as some of the api's are based on dx9

Again, who cares? If MS has the power to lock Linux out of OpenGL 2.0 (through patents?), then open source will just not use it and instead evolve OpenGL 1.* in a different direction.

Re:who cares about "the lead"?

[g4dget]

The OpenGL 1.* which is going to be supported by what hardware in five years?

By whatever well-documented graphics hardware there is going to be in five years.

It's a losing battle. Just like with DRM enabled hardware. Open Source operating systems can only be run on pre-DRM hardware which will become obsolete in a few years.

Come on, wake up. The world doesn't work according to Bill Gates's pipe dreams. First of all, DRM-enabled hardware doesn't exclude open source software: you can either run it without DRM, or you can sign it. Even if it did, there is going to be plenty of non-DRM hardware going to be out there.

Open source is going to be here decades from now. I wouldn't be so sure about Microsoft, however.

Re:mesa sucks compared to dx 9

[k98sven]

Microsoft will have the good will to support those prior interfaces - do Age of Empires or Age of Kings ring a bell? They will not stop supporting their prior games in the foreseeable future.

Good, 'cos I'm having some problems trying to run GORILLAS.BAS on WinXP.

Still, I agree, MS is not going to shoot themselves in the foot
by breaking backward compatibility if they don't have to.

Yes, OpenGL is open, but do you think anyone cares?
The computer games industry is like any other business, it is run by economics.

The kids want games with bleeding-edge 3d, and unless you're name is John Carmack, the industry is not going to support anything else than DirectX.

Now OpenGL 2.0 has the chance to turn the tables;
If GL2 can equal DirectX,
the game industry will use it,
if not for other reasons than economic reasons. Why? One word: Portability.

Porting to the Mac (and maybe even linux) is no problem with OpenGL, but if your code depends on DirectX: forget about it.

Now which game developer would knowingly limit themselves to the MS platform, if they had an equal alternative?

Which game producer would not want to be able to release a Mac or Linux port, at little extra cost?

Re:mesa sucks compared to dx 9

[Anonymous Coward]

Direct3D STILL sucks for scientific visualisation (still favours texture pushing over high-poly-count), and is STILL Windoze-only (Wine excepted). OpenGL sucks less, and is not bound to C++ (Well, o.k. DirectX is theoretically COM, but I defy you to program it seriously in anything other than MS-bastardised C++)

Scientists tend to use grown-up OSes (i.e. no Windoze) and code in Fortran 95 or HPF, pure C or occasionally Lisp - all languages with OpenGL bindings.

You can learn OpenGL+SDL basics in an afternoon, and have flocks of teapots flying across your screen the following morning. Just beginning to learn DirectX and Direct3D means taking on board all the bizarro-world Microsoftian "C++" and COM cruft.

OpenGL's going to be around for some time.

Now, it is inappropriate for hardware raytracing cards, but us people in the scientific graphics community (and movie-making-community) are only getting to play with them now, don't expect them to trickle down to the gaming market for a while yet.

Re:Mesa is not hardware accelerated

[0x0d0a]

Mesa is not hardware accelerated

You mean Mesa's software driver is not hardware accelerated. Take a look at the Mesa FAQ, point 1.2.