OpenGL 3.0
category: general [glöplog]
As some of you may know today the OpenGL 3.0 specification was anounced yesterday changing the API almost completely.
http://www.khronos.org/opengl
Your thoughts.
http://www.khronos.org/opengl
Your thoughts.
Ok, got it. You read slashdot.
Pera's thread is better, he did a full copy + paste!
and links! don't forget the links!!
last time they tried this, someone said "oh wait, does that mean we have to completely rewrite the drivers, including all the debugging, and we do it just to make the world a better place, with no commercial benefit at all?". and then they burned the papers and added some lame extensions.
i'm with chaos on this. allthough i know that makes me sound more prefessional than i am :D
nevertheless, TPB: pounding the future again and again! vote for our nvision entry. already.
nevertheless, TPB: pounding the future again and again! vote for our nvision entry. already.
niels: switch off your internet! :)
Where is the new objectmodel they promised us? I am a bit disappointed, especially considering how long it has been delayed, and considering that the board has not told developers ANYTHING in the past year or so.
Hasn't OpenGL been completely obsolete for years now anyway? Nobody in gamedev is using it - there's GL ES on 1.5 platforms, but no relevant gaming platform (computers, consoles AND handhelds) is using the "normal" GL.
No miracle they didn't redesign the API - apart from a few special cases (most notably mac and Linux nerds) nobody would care. Also very much what Chaos said.
No miracle they didn't redesign the API - apart from a few special cases (most notably mac and Linux nerds) nobody would care. Also very much what Chaos said.
Quote:
ast time they tried this, someone said "oh wait, does that mean we have to completely rewrite the drivers, including all the debugging, and we do it just to make the world a better place, with no commercial benefit at all?". and then they burned the papers and added some lame extensions.
this all of course because is a) respect you guys for being freat friends
Stot 'snel'?" hehe...
kb, just wondering; never looked into it really - does that mean that no modern games run on macs? dx being windows only and all?
skrebbel: EA recently ported a few games. and by ported I mean they used Transgaming's Wine variant called Cider. so uh, woohoo?
yeah, there aren't a lot of Mac games, and I doubt that'll ever change. good thing you can install Windows nowadays on the things. personally I play with my 360, but I can see why people want to play games on computers.
now, if Apple only used decent graphics chips on their cheaper computers too.. :)
yeah, there aren't a lot of Mac games, and I doubt that'll ever change. good thing you can install Windows nowadays on the things. personally I play with my 360, but I can see why people want to play games on computers.
now, if Apple only used decent graphics chips on their cheaper computers too.. :)
kb: that might be more and more true in regard of gaming, but think about other applications such as the simulation industry where this might have a deeper impact.... speaking of money, this one could be even bigger.
I love it how no-one even knows what gl3.0 is, and that even the esteemed farbrausch-people fall into the pit of dooming the api with no knowledge of the actual changes :p
GL3.0 is basically deprecating all the stuff that was hard to write drivers for, making it optional. Like DisplayLists. And moving some of the widely adopted extensions into the core. There is also a bit of porting features from dx10 to gl3, like state-objects to help batching of state-changes.
But all in all the driver-rewriting should be minimal... and absolutely minuscule compared to e.g. dx9->dx10.
GL3.0 is basically deprecating all the stuff that was hard to write drivers for, making it optional. Like DisplayLists. And moving some of the widely adopted extensions into the core. There is also a bit of porting features from dx10 to gl3, like state-objects to help batching of state-changes.
But all in all the driver-rewriting should be minimal... and absolutely minuscule compared to e.g. dx9->dx10.
There's way less games on mac than pc (just market share should be enough to know that), but it's actually not that bad. A lot of the top games get either released at the same time as pc or ported later, plus there are a few mac-only games. I'd say it's good enough for a casual gamer. Not good enough for a serious computer gamer though, especially as quite a few of the best games never get released at all.
That kind of doesn't matter, because you can just boot into windows, but there's still a problem there. For a serious gamer, you want a box that can be upgraded with the latest gpu now and then, on mac that limits you to a mac pro, which is kind of overkill. And even then, video cards on mac tend to lag behind the pc. The current high-end card for a mac pro seems to be an 8800gt... nice, but actually 2 generations old now.
I think most of that comes down to market share though.. nvidia and ati will make a LOT more money from the pc market, so that's where they concentrate their work. Same goes for the games companies. It's changing though, mac market share is growing fast and windows is starting to shrink. If that continues, in a few years attention will start to shift, and opengl for games will be a lot more important again.
That kind of doesn't matter, because you can just boot into windows, but there's still a problem there. For a serious gamer, you want a box that can be upgraded with the latest gpu now and then, on mac that limits you to a mac pro, which is kind of overkill. And even then, video cards on mac tend to lag behind the pc. The current high-end card for a mac pro seems to be an 8800gt... nice, but actually 2 generations old now.
I think most of that comes down to market share though.. nvidia and ati will make a LOT more money from the pc market, so that's where they concentrate their work. Same goes for the games companies. It's changing though, mac market share is growing fast and windows is starting to shrink. If that continues, in a few years attention will start to shift, and opengl for games will be a lot more important again.
I really feel betrayed.. I was looking forward to the new API.
Now all what is left is for one powerful group to push the new API disregarding the ARB...
Now all what is left is for one powerful group to push the new API disregarding the ARB...
pro: Yes, but even the CAD/simulation software is very slowly taking a path towards Windows. That's at least from what I've seen when skimming though the threads on opengl.org et al concerning the 3.0 release (there were quite a few posts by people from the CAD industry chiming into the "fuck that, we're going DX" choir).
Thing is - historically it hasn't made sense for them to use Windows, but since XP it's really on par with Unix machines stability- and "professionality"-wise. They basically have two choices:
- if they want state-of-the-art graphics features on a widely installed hardware base, they'll have to ditch their rendering layer and replace it by something eg. Direct3D based. No big deal, given somewhat clean code the choice of 3D API shouldn't amount to much code anyway, and there's lots of money in the industry
- or if they really need the backwards compatibility because their code is a mess of interwoven rendering and functionality stuff that's only been slowly updated for the last 20 years without any regard of meaningful software engineering - again no problem because there's lots of money aka man-months to throw - they can simply continue like before. But then again - what does GL 3.0 bring to the table that makes it necessary to use?
And hornet: You're really making it sound like it was a good thing. Ever thought about going into marketing? ;)
Thing is - historically it hasn't made sense for them to use Windows, but since XP it's really on par with Unix machines stability- and "professionality"-wise. They basically have two choices:
- if they want state-of-the-art graphics features on a widely installed hardware base, they'll have to ditch their rendering layer and replace it by something eg. Direct3D based. No big deal, given somewhat clean code the choice of 3D API shouldn't amount to much code anyway, and there's lots of money in the industry
- or if they really need the backwards compatibility because their code is a mess of interwoven rendering and functionality stuff that's only been slowly updated for the last 20 years without any regard of meaningful software engineering - again no problem because there's lots of money aka man-months to throw - they can simply continue like before. But then again - what does GL 3.0 bring to the table that makes it necessary to use?
And hornet: You're really making it sound like it was a good thing. Ever thought about going into marketing? ;)
kb, i don't know... even though everybody is loving his or hers xp64-graphics-workstation direct-x with xp is a dead horse, shot by ms. and as far as i am aware of no major 3d app developer is seriously considering the jump to vista... plus a lot of tools get linux and macos ports nowadays. feels a bit like the shift from sgi to nt in the mid-90s...
OpenGL died today.
R.I.P. :(
R.I.P. :(
Okay so my impression on this stuff goes like.. (don't take this as a word of an expert though. i'm somewhat allergic to gpu coding so i don't follow it that much)
What was promised:
-no more legacy bullshit
-no more state machine
-lean api to implement so that intel/sis/amd/etc could actually give us drivers
What we got:
-well, the legacy bullshit is deprecated
-well, you can get around the state machine with an extension
-well, the non-deprecated api is leaner and you could stuff some bullshit wrapper for the rest
-hey some shiny new stuff some 500 euro gpus support
..so we "kinda" got it.
What this means in reality:
-nothing.
To succeed with this model change it's vital to get enough momentum and interest and thus:
-the legacy bullshit will continue to exist.
-the extension in question cannot be relied on.
-instead of wrapping the old stuff using the new api, old stuff just gets extended. haphazardly.
-heck, the whole ogl3 is going to be a curiosity for a number of years for the simple reason that it requires new hardware.
What it doesn't mean:
-opengl is dead. as long there are no alternatives it cannot die even if we wanted it to.
What was promised:
-no more legacy bullshit
-no more state machine
-lean api to implement so that intel/sis/amd/etc could actually give us drivers
What we got:
-well, the legacy bullshit is deprecated
-well, you can get around the state machine with an extension
-well, the non-deprecated api is leaner and you could stuff some bullshit wrapper for the rest
-hey some shiny new stuff some 500 euro gpus support
..so we "kinda" got it.
What this means in reality:
-nothing.
To succeed with this model change it's vital to get enough momentum and interest and thus:
-the legacy bullshit will continue to exist.
-the extension in question cannot be relied on.
-instead of wrapping the old stuff using the new api, old stuff just gets extended. haphazardly.
-heck, the whole ogl3 is going to be a curiosity for a number of years for the simple reason that it requires new hardware.
What it doesn't mean:
-opengl is dead. as long there are no alternatives it cannot die even if we wanted it to.
did you really expect anything different? :)
the khronos group is like the united nations - they'll have some big plans, but in the end most of it will be vetoed by russia (/sgi. /whatever).
i mean, see what they did with collada? :)
whereas on the d3d front, microsoft is like the USA - if they dont like something they can just bomb the shit out of it (erm, add or delete it from the api). and if you dont like it, tough. it may not sound too cool, but it gets results and it keeps things in line.
the khronos group is like the united nations - they'll have some big plans, but in the end most of it will be vetoed by russia (/sgi. /whatever).
i mean, see what they did with collada? :)
whereas on the d3d front, microsoft is like the USA - if they dont like something they can just bomb the shit out of it (erm, add or delete it from the api). and if you dont like it, tough. it may not sound too cool, but it gets results and it keeps things in line.
i am from farbrausch and i approve of OpenGL 3.0
and linux!111
Oh, sorry, I was describing what was to be. 216 has it right :´(