Announcement

Collapse
No announcement yet.

Mesa 19.2 Is Just Six Patches Away From Seeing OpenGL 4.6 Support

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • microcode
    replied
    Originally posted by smitty3268 View Post

    I believe one of the major intended benefits is for people who write HLSL shaders and then port them to OpenGL - such as Valve. Supposedly SPIR-V should make that process simpler and more automatable. Though now that Vulkan is taking off I think the impact of apps wanting to support OpenGL (but only version 4.6) won't be that high. Maybe workstation apps will take advantage of it more than games.
    Yeah, I was thinking about that. Just not sure there are many reasons to do it with GL rather than Vulkan, if you're rolling your own brand new compat layer targeting SPIR-V. :- P

    Leave a comment:


  • smitty3268
    replied
    Originally posted by microcode View Post

    Yeah, basically; it also means that if the publisher of the software is feeling lucky, they'll ship only SPIR-V shaders for both a Vulkan and a GL renderer; though it seems to me that platforms with working OpenGL 4.6 drivers will tend to have working Vulkan drivers as well. Maybe it was intended as a transitional feature, to allow people to start using SPIR-V toolchains before implementing a Vulkan renderer.
    I believe one of the major intended benefits is for people who write HLSL shaders and then port them to OpenGL - such as Valve. Supposedly SPIR-V should make that process simpler and more automatable. Though now that Vulkan is taking off I think the impact of apps wanting to support OpenGL (but only version 4.6) won't be that high. Maybe workstation apps will take advantage of it more than games.

    Leave a comment:


  • pal666
    replied
    Originally posted by treba View Post
    Can applications then load spirv shaders instead of glsl, requiering less shader compilation?
    spirv still has to be compiled to native code. so it will be less, but not by much(still every single bit helps)
    Last edited by pal666; 18 July 2019, 03:49 PM.

    Leave a comment:


  • tildearrow
    replied
    Originally posted by uid313 View Post
    I would like to see some cool demos that utilize modern technology such as Vulkan, OpenGL 4.6, tessellation, ray tracing, etc.
    So that we can see what is possible with computers in 2019, and what is possible with Linux.
    There are some. However, they are mostly closed-source.

    Leave a comment:


  • smitty3268
    replied
    Down to 1 patch that actually changes logic, which is causing a regression. The other 5 just update docs and enable the extensions.

    Expect radeonsi to enable it soon, on the NIR backend. I'd expect the TGSI backend to remain the default (at 4.5) for at least the next mesa release though.

    Leave a comment:


  • uid313
    replied
    I would like to see some cool demos that utilize modern technology such as Vulkan, OpenGL 4.6, tessellation, ray tracing, etc.
    So that we can see what is possible with computers in 2019, and what is possible with Linux.

    Leave a comment:


  • nuetzel
    replied
    Originally posted by schmidtbag View Post
    Don't get too , because in a lot of cases, when one brand adds something to Mesa, it's not uncommon for the other to follow shortly after. Being open-source, I'm sure AMD could mooch off of most of the existing code. Besides... not like there's a lot cases where we're all waiting for 4.6 anyway. I personally don't know of anything that depends on it.
    Fore sure, all have noticed this:

    https://mesamatrix.net/
    OpenGL 4.6 - GLSL 4.60
    GL_ARB_gl_spirv and GL_ARB_spirv_extensions

    in progress (Nicolai Hähnle (AMD) and Ian Romanick (Intel))

    So chances are high for Radeonsi, too. Even that Nicolai was heavy involved with Navi bring up (internal stuff lately).

    Leave a comment:


  • skeevy420
    replied
    Originally posted by schmidtbag View Post
    Don't get too , because in a lot of cases, when one brand adds something to Mesa, it's not uncommon for the other to follow shortly after. Being open-source, I'm sure AMD could mooch off of most of the existing code. Besides... not like there's a lot cases where we're all waiting for 4.6 anyway. I personally don't know of anything that depends on it.
    I know. Still, it's a bit of a let-down when you read the fine print.

    Leave a comment:


  • Terrablit
    replied
    It's a pretty big milestone. I've been following this stuff on Mesamatrix back when those drivers were in the early OGL 4 stages. It's great seeing how much green there is on there now. And it's great knowing that we've almost completely caught up on one standard. The mesa team and everyone else in the graphics stack have done an outstanding job.

    Leave a comment:


  • schmidtbag
    replied
    Originally posted by skeevy420 View Post
    Michael, I'm not normally one to give suggestions like this, but you should add "For Intel GPUs" at the end of that headline. I became really then really as an AMD user.
    Don't get too , because in a lot of cases, when one brand adds something to Mesa, it's not uncommon for the other to follow shortly after. Being open-source, I'm sure AMD could mooch off of most of the existing code. Besides... not like there's a lot cases where we're all waiting for 4.6 anyway. I personally don't know of anything that depends on it.

    Leave a comment:

Working...
X