Page 3 of 4 FirstFirst 1234 LastLast
Results 21 to 30 of 34

Thread: Khronos Publishes OpenGL 4.4 Specification

  1. #21
    Join Date
    Sep 2010
    Posts
    716

    Default

    sparse_texture
    and
    bindless_texture

    Come from different vendors. This mean that those extension wont probably divide OGL capable GPUs into OGL versioned camps, but in vendor camps. (Nvidia GPUs implementing second, AMD first..)

    However since those got ARB status now, there MUST be consensus on them among vendors, so we should see big trio to work on inclusion of needed hardware to their gpus.

    Also 4.x and 3.x (to lesser extend) are ment to be executed on "unified" hardware base, so new features relay on the hardware that already is there.

    So it those extensions require new hw, we may see OpenGL 5 specification in "near" future.


    On the side note: at least AMD seam to be unsatisfied with peace MS develop their DX, so maybe, we could see OpenGL 5 first, and the API that pushes for new level of hw features.

  2. #22
    Join Date
    Sep 2011
    Posts
    707

    Default

    Quote Originally Posted by Nille View Post
    Something changed since the last time?
    Found the quote for you (it's from one week ago):
    Quote Originally Posted by uid313 View Post
    Sucks that there Khronos doesn't provide any reference implementation.
    Nothing changed in regards to Khronos, but uid313 seams nicer about it. Hoverer it bet that if Khronos dropped a proprietary reference implementation tomorrow, with a license that does not allow inclusion in Linux distribution, uid313 wound be just as unhappy. Probably what he really wants is for more people to develop on mesa, and his strategy is to shame anyone involved in the graphs business to do so, rather what he should be doing is to learn C and start doing something about it.

  3. #23
    Join Date
    Mar 2012
    Posts
    130

    Default

    Quote Originally Posted by Nille View Post
    Where?

    ....
    Source: http://www.bit-tech.net/news/bits/20...pengl-opencl/1

  4. #24
    Join Date
    Sep 2011
    Posts
    707

    Default

    A conformance tests is just as valuable as reference implementation, if not more so.

  5. #25
    Join Date
    Jul 2008
    Location
    Germany
    Posts
    711

    Default

    Quote Originally Posted by zxy_thf View Post
    Ty, my only Problem with this is not this part.

    requiring that vendors submit their hardware for full certification in order to boast OpenGL 4.4 compatibility.
    An public test suit and Reference implementation brings much more for the developers.

    Quote Originally Posted by AJenbo View Post
    A conformance tests is just as valuable as reference implementation, if not more so.
    He has only some test cases and you cant compare results. E.g 3 Vendors and each of them has make the conformance test without a problem, but my program runs only on one of them. With a public reference implementation i can test by myself and see how wrong.
    Last edited by Nille; 07-23-2013 at 10:48 AM.

  6. #26
    Join Date
    Feb 2010
    Posts
    39

    Default

    Quote Originally Posted by przemoli View Post
    sparse_texture
    and
    bindless_texture

    Come from different vendors. This mean that those extension wont probably divide OGL capable GPUs into OGL versioned camps, but in vendor camps. (Nvidia GPUs implementing second, AMD first..)
    ...
    So it those extensions require new hw, we may see OpenGL 5 specification in "near" future.
    The new Beta OpenGL 4.4 Driver from Nvidia already features ARB_sparse_texture (all OpenGL 4 capable hardware) and ARB_bindless_texture (Geforce 6xx+ hardware), so no "division in vendor camps".
    AMD is (as usually) slower to adopt newest OpenGL versions (they are still working on full 4.3 support), but I have no doubt that (one fine day) they will not only support OpenGL 4.4, but also ARB_sparse_texture (already having AMD_sparse_texture) and even ARB_bindless_texture, because why should it be ARB approved if only one of the main companys will support it.
    About what hardware level is needed for the new extensions/OpenGL 4.4, here are some clues:
    https://developer.nvidia.com/opengl-driver

  7. #27
    Join Date
    Sep 2011
    Posts
    707

    Default

    Quote Originally Posted by Nille View Post
    He has only some test cases and you cant compare results. E.g 3 Vendors and each of them has make the conformance test without a problem, but my program runs only on one of them. With a public reference implementation i can test by myself and see how wrong.
    I don't see why you wouldn't be able to compare results. I also see any where that the test aren't publicly available. Just that you have to pass the test to get certified. If that is not the case it is a shame as it would make sense to make the test available so that vendors can work on passing them, saving every body time.

    A reference implementation only becomes relevant when the specification is ambiguous and the tests doesn't cover the issue, but even reference implementations can have bugs.

    But none of them would guaranty that vendors chose to follow any of it any way.

  8. #28
    Join Date
    Jul 2008
    Location
    Germany
    Posts
    711

    Default

    Quote Originally Posted by AJenbo View Post
    I don't see why you wouldn't be able to compare results.
    of course i can compare but i cant blame someone because i dont know how is wrong

    Quote Originally Posted by AJenbo View Post
    I also see any where that the test aren't publicly available.
    i guess that because of this "The group also announced the first set of formal OpenGL conformance tests since OpenGL 2.0, requiring that vendors submit their hardware for full certification in order to boast OpenGL 4.4 compatibility."

    In the past the pre 2.0 Tests are also not Public.

    Quote Originally Posted by AJenbo View Post
    A reference implementation only becomes relevant when the specification is ambiguous and the tests doesn't cover the issue, but even reference implementations can have bugs.
    Read some of the Specifications and you Beg for and reference implementation. And, for me the most important part, you has a Implementation where you can test your Stuff.

  9. #29
    Join Date
    Feb 2010
    Posts
    39

    Default

    Because there are sometimes folks from AMD, Intel or Nvidia here: are those formal OpenGL conformance tests based on Piglit?
    If yes, me thinks Piglit should be allowed a more beautiful/up-to-date Homepage with Conformance test results etc.

  10. #30
    Join Date
    Sep 2011
    Posts
    707

    Default

    Quote Originally Posted by Nille View Post
    of course i can compare but i cant blame someone because i dont know how is wrong
    The test would show who is doing it wrong.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •