Just remember, PETA VP Mary Beth Sweetland is a diabetic and needs pig insulin. This little fact can win just about every NAVS and PETA debate you'll ever have to have.
Also, pronounce it "pay-tah", and when they correct you, remind them of a pita, and the delicious meat fillings they usually come with.
Actually, we're fortunate to be able to reuse most of our code on lots of GPUs -- there are some bits that apply from r100 to r700. The "fun" is in the sheer complexity of the hardware, the inability of the hardware to cope with incorrect programming, and the lack of documentation, manpower, and testing available to assist us.
Oh, and hardware bugs. You don't wanna know how many there are. Really. Try getting an RS480, or RC410, to do 3D. It ain't fun.
Any r600 acceleration code *should* work with only minor tweaks on Evergreen (r800). The biggest changes are supposedly in GPGPU-land; r800 supports a lot more shader instructions than r700 or r600.
I don't have one of these yet, but I'm sure Cooper and Richard, the AMD 3D devs, are furiously coding away to make stuff run.
Stuff's really starting to come together. With r300-gallium, I'm almost able to play Openarena reasonably, and my next target is probably Team Fortress 2 because I love torching Spies. r600-gallium is starting to gather attention and we might really dig in on that front in a month or so.
Well, that's a driver problem and not an app problem, and moreover, it's already been solved. Even Mesa already has a fixed-function->shader setup, and Gallium is fully shaderful.
I'm talking about hardware already deployed, not new hardware. All the new stuff coming out is GL 3.1-capable, sure. (Except the boards being sold with old IGPs, of course.)
It is, unfortunately, frustrating to write performant GL these days. While we would like to think that you can do what you want in GL, and the driver will magically make it go fast, the fact remains that some ways of drawing are slow and some are fast. Games are written for specific cards, and drivers get app-specific code. It's a mess, but a necessary mess.
The other problem is that the majority of available hardware and drivers don't support GL 3.x. No open-source driver does, and in fact most Intel, Radeon, and nVidia hardware already in use can only do GL 2.x.
The Red Book is doing what it should. It's providing a transitional viewpoint for writing apps that will work with GL 1.5-2.1 stacks, discussing both the fixed-function and shaderful pipelines. It talks about the maths needed to make things happen, and shows both the old and new ways of accelerating those maths on hardware.
There's a reason that we recommend Red Book to people that are coming in and wanting to write drivers. Well, okay, two reasons. It's available online, for free. Maybe it's a slightly older version, but it's still a good primer. (http://fly.cc.fer.hr/~unreal/theredbook/ is a fairly readable version, but you can find others out there too.)
Oh, look, we already did. We put two memory managers in the kernel, one for discrete chipsets and one for IGP chipsets. We added a system for lockless submission of commands to the GPU, and a series of checks to prevent GPUs from being easily hardlocked.
The reason you're being modded Funny, I hope, is because 3D drivers (and actually any kind of acceleration for GPUs) is a userland thing by tradition and convention. The nouveau project does sponsor development of a few Mesa/Gallium drivers, none of which are yet production-quality, but it has nothing to do with the kernel part.
Well, not overtly, but consider Apple's market position: They make shiny white boxes that are overpriced and pander to a small segment of the market. They have a fifth of the computer market and are not trying to expand, mostly because their vertically integrated business model makes it difficult to increase manufacturing. The Mac Mini proved that they had a cap on their production, and they cannot sell their OS alone without suffering greatly in their business model.
And Microsoft's position: They hold 80% of the market and cannot change. This isn't a problem because many large segments of customers are businesses that strongly desire an unchanging OS. MS has demonstrated a near-unbelievable commitment to binary compatibility and enterprise support, cementing its position. It hasn't been able to keep a strong grasp on the netbook and desktop market in recent years, though.
Now, where's the cooperation? Simple. Microsoft uses its deep pockets and inertia to continue to push itself as the dominant, common, utilitarian operating system, while Apple continually compares its products favorably to Microsoft's and portrays its systems as being hip, cool, modern, and fun. We've all seen the "I'm a Mac, I'm a PC" commercials, and they're representative of the mentality Apple tries to inspire in its commercials, being the small underdog fighting against the big man. Remember "Think Different?"
One Slashdotter has a mangled Voltaire quote in his sig about Apple and MS, but in my opinion it's backward. "If there were no Microsoft, it would be necessary for Apple to create one." However, this will never happen, because Microsoft's power to endure is ridiculous. Just like IBM wasn't destroyed in the decades prior, Microsoft can't be brought down by hordes of Apple fans, or waves of Linux supporters.
Of course, I'm really just re-analysing the premises of World Domination 201 here, but it's not like anybody here has read it, right?
When open-source video drivers begin to get video acceleration for modern formats, Theora will be first on the list due to Redhat's endorsement. h.264 will not be supported in default builds on many distros.
Additionally, if you're using fglrx, you aren't getting video acceleration (fglrx doesn't support it using the standard APIs) so it doesn't matter which codec you use.
Oh, and Android has Vorbis support out-of-the-box. I don't think we lost, not at all.:3
Oh yeah? Well, content shouldn't be used in place of numbers! Never trust those numbers; always hire hard-working letters instead. Alphabetic supremacist for life!
Oh, and I heard you're a lesbian. All livin' it up on your little island of Lesbos with all the other lesbians, am I right?
Windows keys usually send the Super bucky on Unices, which is not necessarily bad. Emacs users rely on them sometimes. I may not be a fan of Emacs, but I can see the usage.
Since the kernel doesn't really care about them, you can free up some of those Ctrl-Alt combos in your window manager (lookin' at you, kwin) by remapping them to Super combos.
It does kind of suck that they're marked with an MS logo, though.
We're a rather bright spot on the university's record; we are the largest open-source datacenter in the hemisphere, and that causes a lot of donations to come in. Take it from Ed: http://osuosl.org/sites/osuosl.org/files/ed_ray.png Nobody will shut us down.
Your post is roughly fourteen months out-of-date. In the past year, TTM and GEM have both matured and been submitted to the main kernel, providing memory management services to nouveau, radeon, intel, and via. GLX 1.4 support is now advertised server-side for DRI2 stacks.
In Mesa, most of GLSL is now supported by the drivers that can accelerate it, and the actual GLSL hooks are now in place for r600 and i965. Additionally, in Gallium, work is underway to provide GL 2.0+ on i915, i965, r300+, and nv30+ (all GeForces after 2004 or so.)
I do feel the need to nitpick a few things. GL 2.1 is pure lies on many chipsets, including a fair number of nV GPUs, so nvidia is not being exactly honest. Additionally, nV has not always provided drivers for the newest and latest GPUs on the market, causing the agonizingly slow vesa driver to be used instead. Finally, you completely glossed over 2D, which is not surprising, because nouveau has been faster than nvidia in all 2D rendering besides video for a while.
Believe it or not, we are at a point where the graphics situation on open-source operating systems is no longer dire.
Not feeling like replying to the troll posts, but I'll reply to you.
I probably reported the TH08 status; I'm kind of a danmaku fan. It was totally playable on an X1950 on an all-classic setup, but there's been a few regressions since and I bet the performance has dropped a little bit. We had to sacrifice a bit of performance to add stuff like FBOs and DRI2. It should start getting better soon though; we've started paying attention to speed and such so the next couple months should see big speedups.
Nightmare Before Christmas was very pleasant as a 3D film, but that's largely because it was originally 2D and the 3D effect was added in later for a re-release.
Support in the open-source drivers is being written as fast as ATI can verify and declassify docs. Also the r600/r700 3D code should be mostly reusable for these GPUs.
It's "true." We tend to "use" lots of "dialectic" and "fact-based evidence" when coming to our "conclusions" instead of feverishly insisting that "everybody" with an opposing viewpoint is "wrong."
Slashdot Mirror
Just remember, PETA VP Mary Beth Sweetland is a diabetic and needs pig insulin. This little fact can win just about every NAVS and PETA debate you'll ever have to have.
Also, pronounce it "pay-tah", and when they correct you, remind them of a pita, and the delicious meat fillings they usually come with.
The people modding you seem to be unaware that 4chan doesn't have memberships.
Theora is H.263, and the patents it is based on have been promised to the community with a guarantee of no enforcement, licenses, or fees.
No wonder you got modded funny.
Troll harder. Theora is H.263, it can be decoded on H.264 decoders thanks to the inherent programmability of said decoders.
Apple has a vested interest -- they get a slice of the patent pie from Sorenson for MPEG-4 licenses.
Actually, we're fortunate to be able to reuse most of our code on lots of GPUs -- there are some bits that apply from r100 to r700. The "fun" is in the sheer complexity of the hardware, the inability of the hardware to cope with incorrect programming, and the lack of documentation, manpower, and testing available to assist us.
Oh, and hardware bugs. You don't wanna know how many there are. Really. Try getting an RS480, or RC410, to do 3D. It ain't fun.
Some reading between the lines is needed.
Any r600 acceleration code *should* work with only minor tweaks on Evergreen (r800). The biggest changes are supposedly in GPGPU-land; r800 supports a lot more shader instructions than r700 or r600.
I don't have one of these yet, but I'm sure Cooper and Richard, the AMD 3D devs, are furiously coding away to make stuff run.
:3
Stuff's really starting to come together. With r300-gallium, I'm almost able to play Openarena reasonably, and my next target is probably Team Fortress 2 because I love torching Spies. r600-gallium is starting to gather attention and we might really dig in on that front in a month or so.
Thanks for your hard work in porting games.
$ glxinfo | grep OpenGL
OpenGL vendor string: VMware, Inc.
OpenGL renderer string: Gallium 0.3 on softpipe
OpenGL version string: 2.1 Mesa 7.8-devel
OpenGL shading language version string: 1.20
OpenGL vendor string: X.Org R300 Project
OpenGL renderer string: Gallium 0.3 on R580
OpenGL version string: 2.1 Mesa 7.8-devel
OpenGL shading language version string: 1.20
Two GPUs. So fuck you. :3
Well, that's a driver problem and not an app problem, and moreover, it's already been solved. Even Mesa already has a fixed-function->shader setup, and Gallium is fully shaderful.
I'm talking about hardware already deployed, not new hardware. All the new stuff coming out is GL 3.1-capable, sure. (Except the boards being sold with old IGPs, of course.)
It is, unfortunately, frustrating to write performant GL these days. While we would like to think that you can do what you want in GL, and the driver will magically make it go fast, the fact remains that some ways of drawing are slow and some are fast. Games are written for specific cards, and drivers get app-specific code. It's a mess, but a necessary mess.
The other problem is that the majority of available hardware and drivers don't support GL 3.x. No open-source driver does, and in fact most Intel, Radeon, and nVidia hardware already in use can only do GL 2.x.
The Red Book is doing what it should. It's providing a transitional viewpoint for writing apps that will work with GL 1.5-2.1 stacks, discussing both the fixed-function and shaderful pipelines. It talks about the maths needed to make things happen, and shows both the old and new ways of accelerating those maths on hardware.
There's a reason that we recommend Red Book to people that are coming in and wanting to write drivers. Well, okay, two reasons. It's available online, for free. Maybe it's a slightly older version, but it's still a good primer. (http://fly.cc.fer.hr/~unreal/theredbook/ is a fairly readable version, but you can find others out there too.)
Oh, look, we already did. We put two memory managers in the kernel, one for discrete chipsets and one for IGP chipsets. We added a system for lockless submission of commands to the GPU, and a series of checks to prevent GPUs from being easily hardlocked.
The reason you're being modded Funny, I hope, is because 3D drivers (and actually any kind of acceleration for GPUs) is a userland thing by tradition and convention. The nouveau project does sponsor development of a few Mesa/Gallium drivers, none of which are yet production-quality, but it has nothing to do with the kernel part.
Well, not overtly, but consider Apple's market position: They make shiny white boxes that are overpriced and pander to a small segment of the market. They have a fifth of the computer market and are not trying to expand, mostly because their vertically integrated business model makes it difficult to increase manufacturing. The Mac Mini proved that they had a cap on their production, and they cannot sell their OS alone without suffering greatly in their business model.
And Microsoft's position: They hold 80% of the market and cannot change. This isn't a problem because many large segments of customers are businesses that strongly desire an unchanging OS. MS has demonstrated a near-unbelievable commitment to binary compatibility and enterprise support, cementing its position. It hasn't been able to keep a strong grasp on the netbook and desktop market in recent years, though.
Now, where's the cooperation? Simple. Microsoft uses its deep pockets and inertia to continue to push itself as the dominant, common, utilitarian operating system, while Apple continually compares its products favorably to Microsoft's and portrays its systems as being hip, cool, modern, and fun. We've all seen the "I'm a Mac, I'm a PC" commercials, and they're representative of the mentality Apple tries to inspire in its commercials, being the small underdog fighting against the big man. Remember "Think Different?"
One Slashdotter has a mangled Voltaire quote in his sig about Apple and MS, but in my opinion it's backward. "If there were no Microsoft, it would be necessary for Apple to create one." However, this will never happen, because Microsoft's power to endure is ridiculous. Just like IBM wasn't destroyed in the decades prior, Microsoft can't be brought down by hordes of Apple fans, or waves of Linux supporters.
Of course, I'm really just re-analysing the premises of World Domination 201 here, but it's not like anybody here has read it, right?
When open-source video drivers begin to get video acceleration for modern formats, Theora will be first on the list due to Redhat's endorsement. h.264 will not be supported in default builds on many distros.
Additionally, if you're using fglrx, you aren't getting video acceleration (fglrx doesn't support it using the standard APIs) so it doesn't matter which codec you use.
Oh, and Android has Vorbis support out-of-the-box. I don't think we lost, not at all. :3
Actually, accelerating Theora is high on our priority list due to it being one of the few codecs that Redhat can sponsor and ship.
We have no plans to add h.264 acceleration to any GPU driver at the moment, although patches are welcome.
The hardware can decode h.263 (Theora) just as well as it can decode h.264.
Oh yeah? Well, content shouldn't be used in place of numbers! Never trust those numbers; always hire hard-working letters instead. Alphabetic supremacist for life!
Oh, and I heard you're a lesbian. All livin' it up on your little island of Lesbos with all the other lesbians, am I right?
Windows keys usually send the Super bucky on Unices, which is not necessarily bad. Emacs users rely on them sometimes. I may not be a fan of Emacs, but I can see the usage.
Since the kernel doesn't really care about them, you can free up some of those Ctrl-Alt combos in your window manager (lookin' at you, kwin) by remapping them to Super combos.
It does kind of suck that they're marked with an MS logo, though.
We're a rather bright spot on the university's record; we are the largest open-source datacenter in the hemisphere, and that causes a lot of donations to come in. Take it from Ed: http://osuosl.org/sites/osuosl.org/files/ed_ray.png Nobody will shut us down.
Your post is roughly fourteen months out-of-date. In the past year, TTM and GEM have both matured and been submitted to the main kernel, providing memory management services to nouveau, radeon, intel, and via. GLX 1.4 support is now advertised server-side for DRI2 stacks.
In Mesa, most of GLSL is now supported by the drivers that can accelerate it, and the actual GLSL hooks are now in place for r600 and i965. Additionally, in Gallium, work is underway to provide GL 2.0+ on i915, i965, r300+, and nv30+ (all GeForces after 2004 or so.)
I do feel the need to nitpick a few things. GL 2.1 is pure lies on many chipsets, including a fair number of nV GPUs, so nvidia is not being exactly honest. Additionally, nV has not always provided drivers for the newest and latest GPUs on the market, causing the agonizingly slow vesa driver to be used instead. Finally, you completely glossed over 2D, which is not surprising, because nouveau has been faster than nvidia in all 2D rendering besides video for a while.
Believe it or not, we are at a point where the graphics situation on open-source operating systems is no longer dire.
Not feeling like replying to the troll posts, but I'll reply to you.
I probably reported the TH08 status; I'm kind of a danmaku fan. It was totally playable on an X1950 on an all-classic setup, but there's been a few regressions since and I bet the performance has dropped a little bit. We had to sacrifice a bit of performance to add stuff like FBOs and DRI2. It should start getting better soon though; we've started paying attention to speed and such so the next couple months should see big speedups.
Nightmare Before Christmas was very pleasant as a 3D film, but that's largely because it was originally 2D and the 3D effect was added in later for a re-release.
Support in the open-source drivers is being written as fast as ATI can verify and declassify docs. Also the r600/r700 3D code should be mostly reusable for these GPUs.
I have three in my system. :3
It's "true." We tend to "use" lots of "dialectic" and "fact-based evidence" when coming to our "conclusions" instead of feverishly insisting that "everybody" with an opposing viewpoint is "wrong."