Comments on: A brief preview of the new features introduced by OpenGL 3.3 and 4.0 http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/ A technical blog from Daniel Rákos (aka aqnuep) Tue, 07 Sep 2010 21:53:56 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: MaNiAc http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-118 MaNiAc Thu, 18 Mar 2010 11:17:44 +0000 http://rastergrid.com/blog/?p=207#comment-118 Very nice post! Wish i had time to write something similar both in length and quality to my blog :) Very nice post! Wish i had time to write something similar both in length and quality to my blog :)

]]> By: jay http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-116 jay Thu, 18 Mar 2010 02:01:22 +0000 http://rastergrid.com/blog/?p=207#comment-116 Thank you for the summary! And kudos to Khronos Group. I wish we can have offline compilation of shaders in a near future release. Runtime compilation is becoming an issue now that shaders a getting more complex. Thank you for the summary! And kudos to Khronos Group. I wish we can have offline compilation of shaders in a near future release. Runtime compilation is becoming an issue now that shaders a getting more complex.

]]> By: Daniel Rákos http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-115 Daniel Rákos Wed, 17 Mar 2010 13:09:00 +0000 http://rastergrid.com/blog/?p=207#comment-115 Yes, both of you are right. It's just my bug that I would like to see more and more extensions integrated into core OpenGL as earlier AMD/ATI didn't really care about the extensions that are not core and it made extensions less useful. Fortunately this changed as AMD is actively participating in the OpenGL world so an ARB extension can be that powerful now without included in the core specification. Sorry if it seems that I'm not happy with the new OpenGL specification releases as I am. It's just that I don't know the background of some of the decisions made and questions pop up inside me, so that's why I'm kind of complaining about these minor things that need clarification. Yes, both of you are right. It’s just my bug that I would like to see more and more extensions integrated into core OpenGL as earlier AMD/ATI didn’t really care about the extensions that are not core and it made extensions less useful.
Fortunately this changed as AMD is actively participating in the OpenGL world so an ARB extension can be that powerful now without included in the core specification.
Sorry if it seems that I’m not happy with the new OpenGL specification releases as I am. It’s just that I don’t know the background of some of the decisions made and questions pop up inside me, so that’s why I’m kind of complaining about these minor things that need clarification.

]]> By: Kevin Rogovin http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-113 Kevin Rogovin Wed, 17 Mar 2010 06:33:30 +0000 http://rastergrid.com/blog/?p=207#comment-113 ARB_draw_buffers_blend was made 4.0 but not 3.3 because, I think, nVidia's GeForce 8 series and up could not do it. They can enable and disable blending per render target (I think) but not change the blending equation per render target. ARB_draw_buffers_blend was made 4.0 but not 3.3 because, I think, nVidia’s GeForce 8 series and up could not do it. They can enable and disable blending per render target (I think) but not change the blending equation per render target.

]]> By: Alfonse http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-111 Alfonse Tue, 16 Mar 2010 23:26:43 +0000 http://rastergrid.com/blog/?p=207#comment-111 On draw_buffers_blend: It is not core in 3.3 because it is not universal in DX10-class hardware. Basically, it's because NVIDIA doesn't support it ;) Anyway, I don't see why it matters so much, especially with core extensions (draw_buffers_blend is not one). Your code paths won't change since you're using the same APIs for core extensions. So if some 4.0 features can trickle down to 3.x hardware through extensions, what is the problem? On draw_buffers_blend:

It is not core in 3.3 because it is not universal in DX10-class hardware. Basically, it’s because NVIDIA doesn’t support it ;)

Anyway, I don’t see why it matters so much, especially with core extensions (draw_buffers_blend is not one). Your code paths won’t change since you’re using the same APIs for core extensions. So if some 4.0 features can trickle down to 3.x hardware through extensions, what is the problem?

]]> By: Daniel Rákos http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-110 Daniel Rákos Tue, 16 Mar 2010 19:32:15 +0000 http://rastergrid.com/blog/?p=207#comment-110 <blockquote>there are some extensions that are only supported on either ati or nvidia “pre-gl4″ hardware (like transform_feedback2 on nvidia or texture_gather on ati)</blockquote> I was disturbed about your statement that ARB_transform_feedback2 is not supported by AMD's Shader Model 4.0 hardware so I did some research... AMD has confirmed that starting from HD2000 Series their GPUs *have* hardware support for the extension and it will be available in the Catalyst drivers. Also got the information that NVIDIA is the one who has GPUs that does not conform to the requirements of ARB_transform_feedback2, at least their pre GTX260/280 boards, but this later is not confirmed by NVIDIA.

there are some extensions that are only supported on either ati or nvidia “pre-gl4″ hardware (like transform_feedback2 on nvidia or texture_gather on ati)

I was disturbed about your statement that ARB_transform_feedback2 is not supported by AMD’s Shader Model 4.0 hardware so I did some research…
AMD has confirmed that starting from HD2000 Series their GPUs *have* hardware support for the extension and it will be available in the Catalyst drivers.
Also got the information that NVIDIA is the one who has GPUs that does not conform to the requirements of ARB_transform_feedback2, at least their pre GTX260/280 boards, but this later is not confirmed by NVIDIA.

]]> By: Daniel Rákos http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-109 Daniel Rákos Tue, 16 Mar 2010 13:06:48 +0000 http://rastergrid.com/blog/?p=207#comment-109 I agree with you, but just partially. Those extensions that target SM 4.1 can be left to a future OpenGL 3.4 or 3.5 version, however those that are already available on SM 4.0 hardware should be available as general, like transform_feedback2. I'm not 100% sure that transform_feedback2 has hardware support on ATI cards but without hardware support, the driver can still emulate the same functionality with better performance than if this emulation is up to the application (talking about the DrawTransformFeedback support). I agree with you, but just partially. Those extensions that target SM 4.1 can be left to a future OpenGL 3.4 or 3.5 version, however those that are already available on SM 4.0 hardware should be available as general, like transform_feedback2. I’m not 100% sure that transform_feedback2 has hardware support on ATI cards but without hardware support, the driver can still emulate the same functionality with better performance than if this emulation is up to the application (talking about the DrawTransformFeedback support).

]]> By: cmf http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-107 cmf Tue, 16 Mar 2010 05:07:41 +0000 http://rastergrid.com/blog/?p=207#comment-107 very nice and informative summary! :) concerning sm 4.x functionality not making it into opengl 3.x: there are some extensions that are only supported on either ati or nvidia "pre-gl4" hardware (like transform_feedback2 on nvidia or texture_gather on ati). so if all these extensions were made core in "opengl 3.4", there would be no hardware to completely support it. very nice and informative summary! :)

concerning sm 4.x functionality not making it into opengl 3.x:
there are some extensions that are only supported on either ati or nvidia “pre-gl4″ hardware (like transform_feedback2 on nvidia or texture_gather on ati). so if all these extensions were made core in “opengl 3.4″, there would be no hardware to completely support it.

]]> By: Daniel Rákos http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-104 Daniel Rákos Mon, 15 Mar 2010 20:25:49 +0000 http://rastergrid.com/blog/?p=207#comment-104 Thanks for the comment! The funny thing is that I've read and written so many times the word "tessellation" and I've never observed that it should be written with two LL :) My mistake, the typo has been corrected. Thanks for the comment!
The funny thing is that I’ve read and written so many times the word “tessellation” and I’ve never observed that it should be written with two LL :)
My mistake, the typo has been corrected.

]]> By: JeGX http://rastergrid.com/blog/2010/03/a-brief-preview-of-the-new-features-introduced-by-opengl-3-3-and-4-0/comment-page-1/#comment-102 JeGX Mon, 15 Mar 2010 20:18:45 +0000 http://rastergrid.com/blog/?p=207#comment-102 Nice overview Daniel. Thanks! Just a typo bug: tessellation with two LL... Nice overview Daniel. Thanks!
Just a typo bug: tessellation with two LL…

]]>