NVIDIA 367.35 Released, Supports 8K H.265 Video Decoding

[spyboot]

With this update some performance problems I previously had with GW2 (massive fps drops, lower framerates the longer the game runs) disappeared. Only slight fps drops (nothing compared to before) when turning the camera are still noticeable.

[duby229]

Sure, half the commits in their git back then before AMDGPU weren't made by AMD employees but it's totally an exception.

That must be why AMD decided that documentation contributions alone wouldn't work and decided to focus on code instead. How many people whined about documentation and then did nothing when it was released?

It's good thing AMD hired the people that actually did contribute.

[starshipeleven]

That argument doesn't work. There are contributors outside of AMD, like Dave Arlie for example, but that's an exception.

Sure, half the commits in their git back then before AMDGPU weren't made by AMD employees but it's totally an exception.

[duby229]

Nope, it's a matter of target. The drivers for linux are mostly developed for workstation use and gaming (3D power), while all the minor things simply don't make sense to spend developer time in.

That's why opensource drivers tend to fare better in the integration part, as the integration is dealt with by other contributors, not main devs usually.

That argument doesn't work. There are contributors outside of AMD, like Dave Arlie for example, but that's an exception.

[starshipeleven]

That's how nVidia does tho. Decent 3d performance, but slow as hell desktop performance and buggy. I personally think it's a problem with how programmers are given a block of functionality to code and then all the different blocks are abstracted together. It makes real shitty code that really can't be debugged. That's why nVidia drivers and Catalyst drivers have bugs that go unfixed for years.

Nope, it's a matter of target. The drivers for linux are mostly developed for workstation use and gaming (3D power), while all the minor things simply don't make sense to spend developer time in.

That's why opensource drivers tend to fare better in the integration part, as the integration is dealt with by other contributors, not main devs usually.

[bug77]

if nvidia is such a shit why so many here have their graphics in their computers? it's totally out of my comprehension.. I can't imagine you turn on the computer the first time as you crack a chocolate egg and then realise "oh fuck I didn't get AMD radeon".. why in the hell??

Eh, this is more complicated than it seems. For years, Nvidia had the best (not perfect, just better than the alternatives) implementation of OpenGL. So for any serious 3D work, you basically had no choice.
Many of the multi-monitor woes you read about have roots in X, Qt/GTK+ or KDE/Gnome, not necessarily in the drivers. That's why some of these problems are hard to debug and sometimes harder to fix properly (i.e. not by introducing a workaround). Fwiw, it looks like Wayland will be much better in this regard and even the modularization that happened in the back end to add Wayland support improved the situation on X.
Now, even if these problems can be justified and understood, they're still exactly the problems that give pause to anyone coming to Linux from Windows or OS X. They make the whole system look unstable and/or not ready. Until every single one of these problems is squashed, desktop Linux will keep looking like a proof of concept

[horizonbrave]

if nvidia is such a shit why so many here have their graphics in their computers? it's totally out of my comprehension.. I can't imagine you turn on the computer the first time as you crack a chocolate egg and then realise "oh fuck I didn't get AMD radeon".. why in the hell??

[duby229]

So KDE is affected too, good to know. A fix for this is long overdue, but I have little hope of NVidia fixing it soon.

That's how nVidia does tho. Decent 3d performance, but slow as hell desktop performance and buggy. I personally think it's a problem with how programmers are given a block of functionality to code and then all the different blocks are abstracted together. It makes real shitty code that really can't be debugged. That's why nVidia drivers and Catalyst drivers have bugs that go unfixed for years.

[JonathanM]

Will this finally fix the issue in KDE when resuming from suspend and the screen is a graphical mess ?

So KDE is affected too, good to know. A fix for this is long overdue, but I have little hope of NVidia fixing it soon.

[deadite66]

I get a black screen on start and I need to go back to VT1 to back to VT7 for it to show me the desktop again. ;(

Yep: https://devtalk.nvidia.com/default/t...en-with-367-35

yep same for me.