Mesa Devs Discuss Potentially Dropping Non-Gallium Drivers Or Forking Code For Gallium

[phoronix]

Phoronix: Mesa Devs Discuss Potentially Dropping Non-Gallium Drivers Or Forking Code For Gallium

Longtime open-source AMD graphics driver developer Marek Olšák has kicked off a discussion over the possibility in the not too distant future of either dropping non-Gallium3D drivers from Mesa (and moving them off to a maintenance branch or the like) or forking some of Mesa's existing code to allow it to be better optimized for Gallium3D use-cases. Due to raised concerns, other possibilities are also being expressed like simply moving ahead with optimizing the Mesa code-base for Gallium3D at a cost of potentially hitting dead code more often with the classic drivers...

http://www.phoronix.com/scan.php?page=news_item&px=Mesa-Possible-Gallium-Route

[jrch2k8]

I hope they listen to Marek because lets be honest classic drivers get broken really really often and is idiotic let ancient drivers to hold performance for modern actually used hardware. As a matter of fact i had to fork my own PKGBUILD for lcarlier mesa's git repo because a lot of times this year i could not compile mesa because either some legacy driver was broken enough to abort the compilation or simply because thing like LTO broke some legacy driver at linking and i was freaking sick of it.

I honestly think that people that actually have that horribly ancient hardware would be better off with a mesa legacy branch only with legacy drivers that receive fixes only and everyone else with a gallium/nir only mesa branch

[microcode]

Now that we have decent GL loaders, no reason not to have two mesas, one with old DRI drivers, and one with new Gallium drivers. This is a great idea.

[jrch2k8]

Now that we have decent GL loaders, no reason not to have two mesas, one with old DRI drivers, and one with new Gallium drivers. This is a great idea.

yeah and honestly 99.999999% of the mesa master commits won't benefit legacy drivers in any positive way and will more often than not break them. I think having a lean frozen legacy mesa will be way more beneficial for those user outside the eventual Xorg fix and just let libglvnd do its job

[c117152]

hold performance for modern actually used hardware

Look up the hardware surveys. 7-9yr/old CPUs are fairly common.

[uid313]

Is Gallium3D still relevant with Vulkan and SPIR-V?
Now there is OpenGL over Vulkan, Direct3D over Vulkan, etc. Maybe OpenCL over Vulkan? There is now also Vulkan Compute.
There is generic mode setting, there is GLAMOR.
What use does Gallium3D serve?

[Niarbeht]

Look up the hardware surveys. 7-9yr/old CPUs are fairly common.

Right, and I'm sure that people use the graphics silicon on those CPUs.

To watch YouTube videos.

Which, let's admit, isn't exactly the pinnacle of performance-centric use-cases.

If it's become a maintenance burden, it probably needs to be split off.

[SilverBird775]

It would be great for IVB and BYT families to join the crowd under the Iris umbrella. The older generations are hardly more capable of what they already are. Even better, the veterans may then happily live in their own erratum space with their own optimized workarounds without dead ends and stubs. For sure the Gen7 are getting their share of hatred due to hardware bugs and last minute design misses, but still this GPUs are capable and widespread. The term widespread have dominant sense here perhaps. You can find the Gen7 user easily with a potent enough machine and they will float around for many years.

[tildearrow]

Is Gallium3D still relevant with Vulkan and SPIR-V?
Now there is OpenGL over Vulkan, Direct3D over Vulkan, etc. Maybe OpenCL over Vulkan? There is now also Vulkan Compute.
There is generic mode setting, there is GLAMOR.
What use does Gallium3D serve?

Yes, it is until the Whatever to Vulkan layers are as fast as their native counterparts.

Gallium3D is a layer that eases the creation and maintenance of drivers for multiple graphics cards. From the official page:

Compared to existing Linux graphics drivers, Gallium3D will:

• Make drivers smaller and simpler.
  • Current DRI drivers are rather complicated. They're large, contain duplicated code and are burdened with implementing many concepts tightly tied to the OpenGL 1.x/2.x API.
• Model modern graphics hardware.
  • The new driver architecture is an abstraction of modern graphics hardware, rather than an OpenGL->hardware translator. The new driver interface will assume the presence of programmable vertex/fragment shaders and flexible memory objects.
• Support multiple graphics APIs.
  • The reduced OpenGL 3.1+ APIs will be much smaller than OpenGL 1.x/2.x. We'd like a driver model that is API-neutral so that it's not tied to a specific graphics API.
• Support multiple operating systems.
  • Gallium drivers have no OS-specific code (OS-specific code goes into the "winsys/screen" modules) so they're portable to Linux, Windows and other operating systems.

[Serafean]

Right, and I'm sure that people use the graphics silicon on those CPUs.

To watch YouTube videos.

Which, let's admit, isn't exactly the pinnacle of performance-centric use-cases.

If it's become a maintenance burden, it probably needs to be split off.

i965 are 2014 macbooks. Still used in production environments (I know because I'm in this case). Nowadays the hardware renewal is 5+ years.
And as a matter of fact a 10 year old desktop with an r600 GPU is still my main home dev machine... Hardware simply ages well these days.