KDE Plasma Readies Its NVIDIA GBM Support, Fingerprint Authentication Added

[TemplarGR]

With 5600XT I've had crashes, black screens and instability under Windows for five months in a row.

Under Linux I had four major bugs which rendered the card useless for me because at least I needed stability and basic features:

https://bugzilla.kernel.org/show_bug.cgi?id=207881

https://bugzilla.kernel.org/show_bug.cgi?id=207735

https://bugzilla.kernel.org/show_bug.cgi?id=207637

https://bugzilla.kernel.org/show_bug.cgi?id=207441

How much effort have AMD put into resolving them? Close to 0. They are all still open but I've long sold the card. Probably worth closing since "CANT REPRODUCE FOR THE LACK OF HW". AMD and Intel Open Source drivers are a godsend until you show Open Source fans critical long standing bugs.

And now that you've started insulted NVIDIA by calling them "Ns-tia" I've lost interest in anything you say or claim. It just shows that you are driven by raw emotions and arguing intelligently is not your forte. For some reasons you believe AMD and Intel hold you dearly and love you, while the don't give a damn about you or anything you do just like NVIDIA. They only happen to spend quite a lot of money supporting the OS whose fans love to shit on everyone and everything that's not open source. You don't understand one important thing: releasing open source drivers is not an advantage for companies, it's a financial burden. Neither you, nor 99.99% of Phoronix users have never and will never contribute a single line of code to their drivers. Your words and principles are hollow. You're basically showing off only it looks cheap and fake.

Every brand new architecture can have driver problems, Nvidia is not exempt. The real reason Nvidia "newer" architectures seem to not have problems is because they are not really new architectures, but iterations of Fermi architecture. Nvidia hasn't made a really new hardware architecture for a decade. Slapping new shit on top of the Fermi architecture, using die shrinks, and just invent marketing names for their mostly software gimmicks, are not new hardware. RDNA on the other hand was new hardware. Unlike the previous gens which were iterations of GCN and had more stable launches.... You see AMD actually attempts to redesign hardware from time to time, and they don't sell marketing BS to their customers at overinflated prices.

[ngraham]

I think the real question is, "how you spell it?"

NVIDIA?
Nvidia?
nVIDIA?
nVidia?
nvidia?
nViDiA?
enn-vidya?
novideo?

[blacknova]

Every brand new architecture can have driver problems, Nvidia is not exempt. The real reason Nvidia "newer" architectures seem to not have problems is because they are not really new architectures, but iterations of Fermi architecture. Nvidia hasn't made a really new hardware architecture for a decade. Slapping new shit on top of the Fermi architecture, using die shrinks, and just invent marketing names for their mostly software gimmicks, are not new hardware. RDNA on the other hand was new hardware. Unlike the previous gens which were iterations of GCN and had more stable launches.... You see AMD actually attempts to redesign hardware from time to time, and they don't sell marketing BS to their customers at overinflated prices.

Wow... that is a lot of BS.
Fermi to current nvidia arch is the same as GCN 1.0 to RDNA, after all if you squint a bit you can recognize them both. Even AMD said that they progressed their design from GCN to CDNA and RDNA. Where former is more compute oriented and the latter is more general usage oriented.

[loganj]

What are you talking about? Kernel developers don't do anything to Nshitia, Nshitia doesn't play well with the kernel, not the kernel devs fault. Also how come your apu is missing vulkan features? RADV is the same for all supported amdgpu hardware. Unless there is a bug or your apu does not support those in hardware....

i said that "for me" thats how it feels. i'm not saying that they do it on purpose. it just feels like that for me.
also about the missing info ive noticed them when ive tried to run the zink test and i got errors that basic vulkan featurea are missing even though the test works.
also the guys from mesa acknowledge that those features are missing when i reported the error.

so now, do i have the right to say that the APU si missing vulkan features?

[mdedetrich]

2) Nvidia drivers on Windows are trash and have been trash for years. AMD has far better drivers on Windows. If you exclude some Nvidia sponsored games that are purposefully made to work badly on non-nvidia hardware for the initial launch period of the game, AMD are much better for gaming, and tend to mature and get optimized with time even more.

Im sorry but until very recently (ergo 2 years) this is bullshit, even AMD themselves admitted that they had to hire a lot more software/driver engineers recently to improve driver quality (which they could do because they are not on the verge of bankruptcy anymore)

Ironically even recently the Linux OS drivers for AMD graphics cards had less issues than the Windows ones.

[oiaohm]

I think the real question is, "how you spell it?"
NVIDIA?
Nvidia?
nVIDIA?
nVidia?
nvidia?
nViDiA?

Welcome to mess.
Nvidia Corporation is the company registration so that gives you Nvidia.
Strange and symbol followed nVIDIA is the company logo for the Nvidia Corporation that gives you the nVIDIA. To be exactly correct here it should be nVIDIA(yes the italic is important to match the logo not not all places can you do the italic) this should be in reference to the logo.
https://www.nvidia.com/Download/index.aspx The driver download page uses NVIDIA and the registered trademarks of most things by the Nvidia Corporation are uppercase yes so a NVIDIA followed by model of graphics card is correct because that will be the trademark for the graphics card.
Yes the all lower case nvidia is the emails/website name.
So that the first 4 and those are in current usage.
nVidia was used in one set of marketing by Nvidia Corporation in history so unless you are referring to that marketing this one is wrong.

So out of that set of 6 only the nViDiA is the only one that the Nvidia Corporation has never used. 5 out of the 6 could be valid. Yes the nVidia is most likely not valid because people are normally not talking about that one marketing compain.

enn-vidya?
novideo?

The novideo is a more annoyance reference due to how often due to issues the NVIDIA drivers result in no graphical output so its technically wrong.

enn-vidya this is wrong some has attempted to write Nvidia name phonetically and screwed up the correct one is /ɛnˈvɪdiə/ or en-VID-ee-ə Yes the capitals on VID is part of writing it correctly. So all the possible correct ones to be used generally.
Nvidia
nVIDIA
NVIDIA
nvidia
/ɛnˈvɪdiə/
en-VID-ee-ə

Can you now see why I wrote welcome to mess.

[Azrael5]

Open Source drivers, while coveted by a majority of open source lovers, seemingly do not mean anything in particular. I've never understood this argument.

• Do open source drivers mean fewer bugs? No.
• Maybe bug reports are fixed faster? No.
• Maybe they mean the product life increases substantially? Not really, considering the pace of progress and the fact that the most work in Open Source is directed towards contemporary products.
• Maybe end users themselves can fix bugs? Almost never happens - GPU drivers are complicated as hell.
• Maybe there are ... more features? Um, no, quite the opposite: closed source AMD and Intel drivers for Windows are leaps and bounds richer than their Linux counterparts. OpenCL and video codecs support is also much better under Windows.
• Maybe features are added faster? Again, quite the opposite, Windows drivers almost always get the new features first. In fact recently an AMD fan has implemented RTRT support in open source AMD drivers, not AMD employees. AMD has kinda given up on it.

So what is it? Can anyone please enlighten me?

What you forget to mention is that the comparison deals with a commercial operating system at high market share, with non commercial operating systems at very low market share. Reasons to provide better support to the commercial product are superior than to support a non commercial product above all when this last one is not widespread. Linux developers work on commercial market just for servers. The only linux commercial product for end-users is Android and its derivatives dominating the smartphone and Tv market.
I don't know the reason why linux operating system on desktop is so limited, but probably it has been always considered a niche product, for amateur purpose in order to make available a corporation-free item. So, the commercial product is supported for a different purpose which involves much more human, material and financial resources, than linux for desktop market, and it has fast evolution and upgrades as well. Probably Linux desktop operating system should be considered much more as an artisanal and artistic product rather than a standardized industrial product. Any industry makes a serialized number of items for the mass of consumers, the artisan makes few unique items with its imperfections and uniqueness improving by experience. The artisan tends to love is product as it was the parent of its daughter or its son. The industry consider the item based on money. Yes, both the items are operating systems, but what is different is the context and the purpose as well. The comparison must consider these differences.

The problem is that hardware is provided from those corporations generating a sort of dependency, because to attract users an operating system needs to be stable and usable. In my opinion, a compromise could be to consider the corporation support as a kind of free contribute for a free open-sourced product which employs their hardware items.

[Yeayo guy]

My last test of KDE Wayland:

- Can't turn off monitor or the desktop goes away
....

Yes I have the same problem, but Edmunson says it should hopefully fixed in 5.23.2

https://invent.kde.org/plasma/plasma-workspace/-/merge_requests/1125

[chocolate]

Quite a lot of pages to reiterate over "transparency & humility vs. toxic pride" (a matter as old as the Debian manifesto) and "prefer open drivers' integration into the GNU/Linux ecosystem vs. prefer Nvidia's package".

People can be enthusiastic about their experience without being necessarily malevolent in their omission of specific problems concerning the libre ecosystem, although it can happen, therefore some degree of scrutiny is entirely natural and not to be taken personally.
People are entitled to their preference of a proprietary package inside an otherwise open/libre operating system: philosophy around free software is merely a suggestion, and it must remain as such in good faith for it to keep having any validity at all in case of technical shortcomings.

birdie Of course you are appreciated for your bug reports and experience, but from the outside it looks like this discussion is sterile and one-sided. We want to highlight problems in the spirit of transparency, but we don't want to take a dump on (overly?) enthusiastic users every single time. Especially if, in the end, there's no lesson for them to learn. Please note that I'm tagging you specifically because I trust your character and your influence in these discussions, and not for any other reason (meaning I know for a fact you bring something positive to the table and I'd like others to acknowledge that as well). Cheers.

[geearf]

At least with NVIDIA I have an option of using multiple drivers versions with my current kernel. There's no such option for Intel and AMD - they have drivers bolted (since you're a huge fan of the term) to your kernel release and you have zero freedom in choosing what to run. And what if your distro is late with a new kernel release which contains support for new HW or major fixes? You're SoL. NVIDIA? Uninstall the current driver, install the newest one even without rebooting.

Aren't you able to do just that by using the official AMD installer and picking the OSS stack?
It should work about just like nVidia's installer I think.