Slashdot Mirror
OpenGL Version 4.3 Released
An anonymous reader writes "The Khronos Group has released the specification for OpenGL 4.3 at the SIGGRAPH 2012 conference in Los Angeles. New functionality includes: compute shaders that harness GPU parallelism for advanced computation, shader storage buffers, improved debug message output, high quality ETC2 / EAC texture compression as a standard feature, memory security improvements, robustness improvements, texture parameter queries, and more."
The Khronos Group also released the OpenGL for Embedded Systems 3.0 specification, which is backwards-compatible with version 2.0. The new specification includes enhancements to the rendering pipeline, "a new version of the GLSL ES shading language with full support for integer and 32-bit floating point operations," and improved texturing functionality, among other things.
Thank goodness the Khronos Group took over from the old OpenGL Architecture Review Board (ARB). There has been great progress in OpenGL since then, catching up to Direct3D which had come from behind. With this OpenGL we can have this goodness on all desktop (Windows including XP, Linux, Mac, Unix) and mobile computing platforms (iOS, iPad, Android). Personally I'm most looking forward to the improved debug message output - hopefully that should save me some time tearing my hair out trying to resolve my mental model of what is going on vs. the realities/subtleties of GPU programming.
When is apple going to get with the program related to 3D graphics? With Lion, they finally released drivers for OpenGL 3.3. Now, they are currently about 4 generations behind this new release. You would think, that with their success of their devices with fairly nice GPU's, that they would try to court gamers and developers. Let's face a hard truth. The most successful apps past and present are games. I know they want their drivers to be stable and all but they are way behind. I don't understand why they can't work with amd and nvidia on getting some stable driver releases...especially now with retina displays.
Perhaps someone can explain what there thinking is here because I feel like they are missing out on some opportunities.
I always wondered why OpenGL never caught on, until I read this explanation at stackexchange.
Great.. :-( one more version for Mesa to be behind of..
Comment removed based on user account deletion
"Windows is slowly losing relevance" - by peppepz (1311345) on Wednesday August 08, @02:22AM (#40915185)
#1 Most Used/Biggest Marketshare on PC Desktops + Servers combined, & it's "losing relevance"? Then MacOS X + Linux never had it @ all, just based on the numbers, & don't argue with me - as the saying goes, "argue with the numbers": See here, "Read 'em & weep" -> http://en.wikipedia.org/wiki/Usage_share_of_operating_systems
---
Parent said "Windows is slowly losing relevance", the article you refer to shows current market share of operating systems, not change in time, so here are some relevant numbers you can argue with.
From September 2008 to April 2012:
Windows: 90.87% -> 84.13%
Mac: 8.69% -> 14.80%
Linux: 0.41% -> 0.86%
So it seems it is true that Windows is slowly losing relevance. In the same period of time Linux doubled its usage. And I suspect they are not taking into account mobile devices such as cellphones and tablets.
I've used AND created OpenGL screensavers for Windows since Windows 2000, XP, Server 2003 - based on the OpenGL 2.1 standard
If your screensavers look anything like your posts, I'm not interested.
After Windows NT, OpenGL has been a second class citizen there, and after Windows Vista, it became a third class one. It's hardware manufacturers that still provide the independent OpenGL implementations you enjoy on Windows. Microsoft's OpenGL implementation (version 1.1 I think) is a remnant of the NT era, when Windows was entering the workstation market, and that meant it had to support OpenGL.
As a mobile and game developer this is great news and I am excited to test some of this myself. :3
Don't feed trolls.
I'm old enough to remember when discussions on Slashdot were well informed.
When I said that "OpenGL on Windows is a second class citizen", I meant that Microsoft does no longer develop it, document it, promote it, not that you can't have good third-party OpenGL implementations (and games that make good use of it) on Windows. Check the OpenGL documentation on MSDN, it's stuck to OpenGL 1.1, probably dating back to the times when OpenGL was THE high performance graphics API for Windows NT.
Apparently there is now some level of convergence between OpenGL and OpenGL ES, so that a valid OpenGL ES 3.0 application is also valid OpenGL 4.2. Does this also apply to ES 2.0? In other words, is a valid ES 2.0 application also a valid non-ES 4.2 application?
if you have a fairly recent mac chances are the graphics hardware can do ALL the great OpenGL stuff. Thats what makes it even more puzzling. Its not that the hardware on these systems aren't there. There just isn't any drivers. Right now my macbook pro can perform all OpenGL 4.1 commands. But no drivers since Apple has been twiddling their thumbs.
here is something that I came across on arstechnica about graphics improvements with Mountain Lion
A little below halfway down the page:
"However, the improved drivers appear to be working in concert with overall improvements to OS X's graphics subsystem. According one source who spoke to Ars on the condition of anonymity, Apple has significantly altered the architecture of the graphics subsystem in Mountain Lion, cleaning up the interface between OpenGL and drivers in order to implement upcoming support for OpenGL 4.2. OS X currently supports version 3.2, and the lack of support for the latest version shows when comparing performance with Windows"
The sun is too small to go supernova. Lrn2science.
Watch for Penguins, they eat Apples and throw rocks at Windows.
for example, was the high quality texture compression already in ati/nvidia implementations?
that and questions like that were probably on the mind of the guy asking what it matters for joe schlobs - not does opengl matter, but how do these new additions matter and if they matter in a tangible seeable way.
world was created 5 seconds before this post as it is.
The blobs currently don't lag behind because they share a lot of code between os versions.
Nvidia's approach is basically throwing out of the window anything that exist under Linux and recompile their windows code. Although this brings tons of drawback (basically nvidia's drivers don't play nice with anything else, and lot of advanced features are missing - like the Optimus debacle and Linus' big fuck you), it has the advantage that they support their own bells and whistles and so OpenGL 4.3 is already out.
AMD's approach is finding a compromise to make their shared catalyst code play a little bit better with Linux standards by bending and twisting it a little bit. Drawback is that it takes a little bit longer, and can be slightly more buggy, advantage is that you get the latest goodness (OpenGL 4.3 soonish) and the latest Linux goodness (full Xrandr support since ages, for example).
Intel is the exception here: They don't have blobs. They actually pay developers (formerly Tungsten Graphics, now bought up by VMWare) to develop fully open-source drivers. The advantage is that they get the latest Linux features or even played a role in developing them (KMS, TTM/GEM, Gallium3D, etc.). The drawback is that they are limited by Mesa's support (currently Gallium3D is only fully compliant to 3.0 although there is ongoing effort to add support for the necessary extension to pass compliance to the rest of the 3.x serie)
So yeah, currently Mesa is a bottle neck.
For now.
The involvement of Valve into Linux gaming might change this trend.
Valve is having linux developers on their payroll. And they aren't only interested in ports. They are interested in developers with driver knowledge too.
There's an active collaboration with Intel to get Mesa advance, specially regarding extensions that Valve needs. For example, while Mesa's current target is getting the rest of the 3.x-serie compliance, Valve and Intel did collaborate to get some debugging function properly implemented (which are required for 4.x compliance).
If Valve keeps throwing efforts and resource at Linux (which they might given their current struggle against Microsoft over the next Windows version 8 and its built-in store) the pace at which Linux and Mesa in particular progress will speed up.
Intel will also play a bigger role in the future: as their GPU are becoming "well enough" not only for desktop needs but even for basic gaming needs (plays anything which is not Crysis 3 at a decent speed), they are gaining market share, simply because they are the "stuff built-in by default" (mid-range and business mainboards with enough graphic power in the chipset, laptop without a Nvidia MXM module, etc.) and as such and because they are fully opensource, they attract also linux users and thus some developers.
And there's a networking effect in play too. Unlike the binary world where everyone is out on his own, with open-source there are a lot of shared frameworks and facilities (the stuff that Nvidia tries to ignore). Development done by one can benefit the others. (Specially given the modular infrastructure of Gallium3D).
So improvement done to Gallium3D by Intel can also improve the opensource driver of AMD.
And AMD is taking part into the opensource development. Maybe the legal-approval process for their documentation release is moving at glaciation speed, but on the other hand they are trying to help and have even a few opensource developers on their own payroll.
(And last but not least, although Nvidia themselves tend to kick all this community development in the nuts, the 3rd party reverse engineered opensource drivers Nouveau can both benefit from- and contribute to- Gallium3D development).
Even Google throws some resources in from time to time, like helping port older Intel hardware to the newer Gallium3D framework.
We are slowly arriving at a solution where a lot of interested parties are putting resources into development. And said parties include hardware developers themselves (Intel, AMD) and gamin
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
OpenGL is a standard, a specification, an API.
Catalyst, Mesa/Gallium3D, etc. are actual implementation, actual code.
release-wise: Nvidia and AMD have compliant drivers out.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
Way to miss the entire point, buddy.
The fact is they are DECLINING, and lost 6%, mostly to Apple during a *four-year* period.
Meanwhile there is Wii, PS3, Linux, iOS and Android running OpenGL *or* OpenGL ES *or* a subset/superset of these. Thus, yes, OpenGL is very relevant today, and no, it's not any harder than Direct3D. But on the desktop it might very well be too late.
systemd is not an init system. It's a GNU replacement.
Addendum:
Just as an example of the involvement of Intel in speeding up the OpenGL 4.3 / OpenGL ES 3.0, this just in at Phoronix:
http://www.phoronix.com/scan.php?page=news_item&px=MTE1Njg
OpenGL ES 3.0 announced by Intel for the opensource drivers for their next generation of graphic cards (currently still alpha quality), and should became available during the first quarter of next year.
Okay, it's not the full OpenGL 4.3 specs, it's only the ES 3.0 subset of it, there still some work to go. But, thanks to the modular architecture of Mesa/Gallium3D this front end is insta-magically available to any back-end exposing the necessary hardware features.
That means that potentially 2013 will not only see support for OpenGL ES 3.0 for Intel HD, but also for older Intel hardware (thanks to Google) or AMD hardware (thanks to their opensource initiative) or Nvidia hardware (no thanks to Nvidia. Fuck you ((c) by Linus), but thanks to the countless reverse engineers), as long as said developers expose the corresponding hardware functionality in their drivers. (And same for the embed chips running Gallium3D based opensource drivers, like Mali on Lima, Freedno on Qualcomm, etc.)
Pour enough resources on it, and full Open GL 4.3 becomes realistic within 2013. (Intel/VMware, Valve, AMD, Google, ... I'm looking at you !)
Now as a scientist, I would enjoy to see some love given to the OpenCL front end too.
(It's not lagging that much behind API-wise because OpenCL is a rather recent standard too. But it's not stable enough and some back-end aren't yet able to use it.
On the other hand, this has some efforts paid by Google and even Pathscale).
And I would guess that a few Linux Gamers would like to see the DX10/DX11 front-end some more serious development beyond the proof of concept stage.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]