Artem Tashkinov: Independent Hardware Vendors Hate Linux

[-MacNuke-]

Sure they hate it. They want to release a quickly hacked driver as closed source with poor quality, never do any updates or fixes and drop support as fast as they can so you have to buy new hardware.

Look at the Intel drivers on Windows... no Vulkan and no OpenGL 4.5 on Haswell. Rendering quality is pure shit. The Vulkan driver on Skylake is buggy as hell.

No OpenGL >4.1 whatsoever on macOS, quality is shit. Apple drops older hardware by not shipping any drivers for older hardware.

OpenGL quality with AMD closed source drivers is horrible.

Yeah. Tell me again that closed source drivers deliver better quality...

[GruenSein]

Sure they hate it. They want to release a quickly hacked driver as closed source with poor quality, never do any updates or fixes and drop support as fast as they can so you have to buy new hardware.

Look at the Intel drivers on Windows... no Vulkan and no OpenGL 4.5 on Haswell. Rendering quality is pure shit. The Vulkan driver on Skylake is buggy as hell.

No OpenGL >4.1 whatsoever on macOS, quality is shit. Apple drops older hardware by not shipping any drivers for older hardware.

OpenGL quality with AMD closed source drivers is horrible.

Yeah. Tell me again that closed source drivers deliver better quality...

To be fair, it is a very recent development that the open source drivers for AMD for example have started to be competitive. And we are all whitness to the growing pains this entails. I am not saying that this isn't the way to go but shit drivers have very little to do with being open source or not. At the moment, AMD is burdened by maintaining two drivers on linux. Nobody knows if the progress of their open driver will translate to other platforms or if it will simply be an additional effort. Also, Nvidia is frequently praised for their oh-so-great closed source drivers. Driver quality has little to do with being open source.

[Luke_Wolf]

My response to this "article":



It's just this week's episode of Birdie lies about shit. Really nothing to see here. Move along. What's next Michael are you going to write articles about what AboutTheBSDs writes?

[rabcor]

" ... will never succeed ... " is already killing any serious discussion on this topic because the Linux-friendly community will go ballistic since his statement is definitely wrong in the generality he puts it in. The desktop is another story completely and the points he's mentioning are worth considering at the very least. Anyway, I'll get the popcorn..

You're wrong in assuming that, it sounds to me like he's feeling desperation, "as long as these issues I am pointing out persist, linux will not succeed" is what he means. You're the one misinterpreting his intentions, what he's really trying to do is convince you by adding weight to his words. "Never" is a pretty heavy statement to support after all.

In a nutshell a guy noticed a problem he finds super serious, and he wants other people to solve it.

[-MacNuke-]

Driver quality has little to do with being open source.

Sure. But at least someone can do something to make it better (look at radv). Windows and macOS are showing exactly what happens with closed source drivers. Nobody fixes anything long-term. I know that it does not happen with every Open Source driver but you can't say "this driver there does not work, so every open source driver is useless".

[R00KIE]

Looks more like a rant from someone used to how things work in the windows world than anything else. It has some truth in it but never addresses the elephant in the room.

Let's rewrite it a little bit, independent hardware vendors hate to support the Linux kernel for many reasons:

- They would be put to shame by the code quality.

- They are too lazy to test with one LTS release and mainline.

- If there are no spec sheets how do they write drivers for windows in the first place? Copy paste of some shoddy code that comes from who known where? DRM isn't stopping AMD and Intel from making drivers available for their hardware, just put a disclaimer saying features X, Y, Z are not supported on linux.

- Again, the claimed massive changes haven't stopped AMD, Intel and many others, even Broadcom is joining the party. The kernels used by redhat have functionality bug fixes and most importantly security fixes, yes that's a quality problem, linux still as some catch up to do because we all know windows is better and does not have security or functionality bugs.

- Considering the amount of people working on linux I'm surprised it works as well as it does, while true that kernel devs many times don't have access to the hardware they will do their best to help someone willing to do their part. The problem here is the mindset, with windows you complain that it doesn't work and if enough people complain maybe the vendor will bother to cobble up a fix. On linux you bisect and report the commit that breaks things, you don't need to know programming to do that, stop whining and do something to help fix your problem.

[Delgarde]

Not being a developper myself, I was surprised to see the statement that the Linux kernel doesn't have a stable API/ABI, seeing how Linus was using strong language to state that "WE DO NOT BREAK USERSPACE!" (https://lkml.org/lkml/2012/12/23/75). Are these two different things or is one of the two statements wrong?

The key word in the Linus quote is "userspace". The interface between kernel and non-kernel code is considered stable, and if kernel developers break that interface, Linux will yell at them. A lot.

The same is not true *within* the kernel - and that includes modules, which despite being dynamically-loaded, are still part of the kernel. And that's not much of a problem for drivers developed within the kernel tree, because developers are usually pretty good about not breaking other parts of the tree... probably because unnecessarily creating work for other kernel developers will also result in Linus yelling at them. But it *is* a problem for drivers developed outside the kernel tree - the attitude tends to be that if they're not doing things the proper way, they can suffer the consequences... and Linux will yell at them instead (see NVidia).

[bug77]

The guy has some points, but none of them make Linux more or less of a success. Success is relative (is ruling the server world success? is completely missing the smartphone craze success?), but what this guy calls "problems" are actually conscious choices. Any engineer worth their pay check knows that's what you do to push things forward. It's as simple as that.

[swoorup]

The guy has a point: stable API/ABI for drivers. When a hardware vendor writes a driver, why would they want to maintain it, after every few kernel releases.

[Joker_]

The guy has a point: stable API/ABI for drivers. When a hardware vendor writes a driver, why would they want to maintain it, after every few kernel releases.

They don't need to. I believe it is common practice that when API changes, those responsible for the change also update drivers that depend on that API. As long as the manufacturer pushes their driver code upstream, there will be others that are willing to maintain it. If they insist keeping it to themselves, they, of course, have to maintain it themselves. But that is simple choice they have made.

Also, see https://github.com/torvalds/linux/bl...i-nonsense.rst