NVIDIA To Meet With Wayland, Linux Kernel Developers To Discuss GBM vs. Streams

[SaucyJack]

Nvidia dares to use a Khronos defined standard? Better post a Linus middle finger. https://regmedia.co.uk/2012/06/18/torvalds_bird.jpg

Phoronix forum's-ing like a champ. #DoingPhoronixRight

Wow you're dumb. EGL streams is only a Khronos standard because Khronos is headed by Nvidia.

Honestly this is pretty disgusting. Nvidia 'we're not going to implement GBM, so instead we are going to wait months to assemble a bunch of people in person to bully the Wayland devs into compliance with what we want.'

[DanL]

Ironically, Intel's open source drivers are regularly trashed on here for being tied to Mesa instead of the LLVMpipe "standard" (even though Mesa and Intel's open source driver existed first).

I guess when you are talking about LLVMPipe, you mean Gallium3D?

So basically it looks like you should always use the Mesa "standard" except for when the crowd arbitrarily decides that Mesa is bad and you shouldn't use it.

That makes no sense whatsoever. You're using the word Mesa in two different ways. Older style Mesa 3D driver vs. Gallium3D driver is not what we are talking about here. The Gallium3D drivers/infrastructure are still a part of the Mesa project. So here we are talking about the Mesa project and its drivers choosing to use their own implementation as opposed to a Khronos standard.

[Gusar]

There is precisely one implementation of EGLStreams, a closed source one, released by Nvidia in their driver only a few months ago. All Wayland compositors use GBM, which has been developed in the open for years already. So what are compositor devs to do now, spend significant amounts of time to add EGLStreams support, and then be required to maintain two backends even though one of those backends will only be used by a single closed source driver, just because EGLStreams is a "standard"?

[bearoso]

I suspect there is some shortcoming in Nvidia drivers or hardware that EGLStreams is working around.

My experience is that Nvidia drivers tend to do a lot of excess output buffering, and is unable to provide the level of synchronization at low latency GBM needs, so this is an attempt to hide it. Thus the question is whether or not GBM is actually possible with Nvidia's existing driver stack.

[Hi-Angel]

Can somebody provide a good technical discussion of the pros/cons of GBM vs. EGL Streams?
So far all I have heard is name-calling against Nvidia for supposedly not following "standards" since they aren't using GBM but from what little I can tell, EGL streams are a cross-platform documented standard that's part of EGL and that goes all the way back to 2012, well before Wayland was a viable project: https://www.khronos.org/registry/egl...KHR_stream.txt

In addition to what other said, the technical discussion itself is linked at the linked articles, e.g. here and there, and probably there was more. In short, what NVidia wants would be a monstrous developer effort with little, if at all, gains.

[log0]

...project and its drivers choosing to use their own implementation as opposed to a Khronos standard.

Jeez, there is no Khronos standard that allows you to implement Wayland.

[SystemCrasher]

This is basically Mesa vs EGL. Wayland is tied to Mesa which won't work on anything that's not Linux

Wait, I think I've seen some VMWare ppl using MESA even on Windows. Can you imagine it, these nuts are using MSVS to build at least some parts, yikes! It gone so badly they can't even use C99 in their part of code, can you imagine THAT?

[SystemCrasher]

nVidia likes shims. Now they get to shim their shim.

Problem solved.

The real problem is the fact nvidia failed to get the idea they have to be part of the process to get their voice heard. They could mumble about khronos standards or whatever, but since they are not part of processes around Linux graphics the only option I could imagine is nvidia coding this crap themselves and maintaining it "forever". Other devs around aren't huge fans of driver blobs, so nvidia would have hard time persuading other devs to write and maintain code for their stinky blob.

[asdfblah]

I don't get it. Doesn't this means Mesa devs are giving up to proprietary software? If so, why?

[ngkaho1234]

The best example I can bring for this issue is the side where the drivers seat to drive the car.

Imagine that the whole world drives the car with the driving seat being on the left side, then USA/EU decides that their cars will have the driving seat on the right (like in UK) in the following years, and if you wanna drive a car from them, you have to adequate.

The whole world unite and say "That's not how our cities were planned, the drivers seat on the left and the cars stand on the right side of the lane", but USA/EU says "We are a big economy, we think drivers on the right, cars on the left are the best way, the whole world didn't realize that this standard is the better one."

What happens next, is that USA/EU (Nvidia is this case), will try to push their way because everyone want a car from them, or the rest of the world will heard complains about the USA/EU cars can drive on their country because of politics

Well, US/Many EU countries will break the legacy position of steering wheel, which is not exactly the case to NVIDIA (iirv nothing major will be broken for NVIDIA to adopt EGL stream rather than GBM)