cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,060
According to Michael Larabel of Phoronix, FreeSync support (or VESA Adaptive-Sync / HDMI VRR) is going to be merged with the upcoming Linux 4.21 kernel cycle. This will enable full FreeSync support for Linux gamers who own an AMD GPU! Mr Larabel also found new AMD Zen linux-firmware, and a workaround for the AMD Radeon RX 590 under Linux.

Assuming no major last minute issues, the code should then be merged into the Linux 4.21 mainline kernel when its merge window opens at the very end of December or early January. Linux 4.21 should then be released in March for those that may be planning a new monitor purchase or upgrade to one of these capable displays.
 
I like how they list the tentative March release of the kernel as some sort of benchmark- they do know that real distros that people use beyond Arch and Manjaro are going to be way behind that?

Also, what's with the brouhaha about this going into the kernel? The support is already there in the kernel module package, right?
 
I like how they list the tentative March release of the kernel as some sort of benchmark- they do know that real distros that people use beyond Arch and Manjaro are going to be way behind that?

Also, what's with the brouhaha about this going into the kernel? The support is already there in the kernel module package, right?

I means you can use 100% free open drivers and use freesync. AMDs open source stack has had a lot of work poured into it. To the point where its faster in many cases then the closed source AMD kernel module option. In comparison Nvidia as an example provides close to zero support to their free in kernel developers... they provide just enough support that Linux will boot allowing people to bolt on the Nvidia kernel module.

On the kernel stuff... IMO everyone should be on manjaro.

Having said that... all the major distros have fairly easy way to run the latest verisons of the kernel. Yes their "stable" versions may ship with an older LTS... but anyone that is smart enough to want to turn free sync on (as you point out with free drivers) can handle switching to a newer kernel line.
 
Don't forget that Freesync is supported by Intel as well. If they get the Linux drivers ready it means a lot of devices with Freesync.
 
Don't forget that Freesync is supported by Intel as well. If they get the Linux drivers ready it means a lot of devices with Freesync.

This won't happen until Intel rolls a new graphics generation. Current Intel iGPUs are still Skylake (or Haswell?) generation.
 
I means you can use 100% free open drivers and use freesync.

I understand that some are theologically motivated here, but that doesn't seem to be a big deal for Linux gaming. You're going to want the most up-to-date drivers regardless, which is why I don't mind Nvidia's stance- they have the best hardware and the best drivers.
 
I understand that some are theologically motivated here, but that doesn't seem to be a big deal for Linux gaming. You're going to want the most up-to-date drivers regardless, which is why I don't mind Nvidia's stance- they have the best hardware and the best drivers.

But there are actual gaming advantages. As Pieter pointed out Freesync is not closed source. Once support for it is in the kernel and in the stack, it can be flipped on for intel... or any other GPU manufacturer. Some day it may matter to an ARM manufacturer using ARM or their own GPU implementations as well.

Their are performance issues with using bolt on kernel modules. Their are issues with constant kernel rebuilds.... sure many major distros pre build nvidia things in alternate kernels. However it menas basically keeping an added kernel in the repos.

Nvidias closed source also slows related development. Nvidia doesn't explain how things work so building universal just works stuff for things like Wayland is next to impossible. Wayland may never be able to completely replace X and that is in large part due to Nvidia. They have used their hardware clout (install base) to push their own proritary standards that hurt performance on everyones hardware but their own. Forcing projects to decide if they want to go ahead with new superior tech and try and maintain 2 code pipelines. For projects the size of something like Wayland or KDE or Gnome ect... its hard enough keeping a core number of high quality developers engaged in a project without creating them completely duplicate sets of work. Recently Nvidia has offered to provide paid support to projects like KDE to support their wayland stuff... but as a matter of principle many of us believe they should tell Nvidia to shove off. That isn't just a ideological stance... creating more code to do the same thing a unique way for X company is both a terrible precedent to set as well as just bad code. Why write 20,000 lines when 10,000 will do. Cleaner more elegant code is both faster and easier to maintain.

I look forward to Intels new GPUs. I don't expect they will take any speed crowns out of the gate. However no matter how much flak you can throw at Intel... their support for open source has always been first class. When they start shipping it should be cool to see both AMD and Intel using almost 100% the same software stacks in Linux. The kernel drivers will differ but all the back end things Mesa the vulkan libraries... the wayland implementations ect will be the same.

I think Linux if nothing else has proven that for large complex bits of software... having multiple companies eyes on the ball improves end users experience. The tide lifts all boats. Nvidias stance is the standard old world corp take on software... why should our work benefit our competition. The same arguments where made in board rooms back in the 70s when things like Postix where contemplated. Bottom line standards are good for consumers... they are good for competition, and ultimately they are good for hardware manufacturers as well. No one company could afford to say develop Linux... and I would say Graphic stacks are much the same way. Nvidia can right now afford to spend supporting graphics stacks cause they are making good bank... but what if Intel stomps them for a couple cycles out of the gate. What is the first area they will make cuts in... Software. Open is good for everyone.
 
Last edited:
But that's still just theory?

That bolt on modules affect performance, etc.- because in theory, they must, but in practice, is the difference measurable?

Beyond that, I have no issue with Nvidia protecting their IP through closed-source drivers so long as said drivers interoperate well and do not hinder performance, and from what I've seen, they work great.

I also have no issue with AMD doing what they can, in this case garnering sentimental support by opening up their code. I wish them luck, I just don't fawn over them through some alignment of ideologies. I'm glad Intel is more or less following suit, but again, they also really have to. I expect my next laptop upgrade to be another Intel machine, but with a 120Hz VRR display. AMD if they can get all of their ducks in a row the way Intel has.

I think Linux if nothing else has proven that for large complex bits of software... having multiple companies eyes on the ball improves end users experience.

I'd argue the complete opposite- Apple being the sole curator of their 'experience' seems to provide the most coherent one, and Windows 10 is still far more consistent and responsive than any of the dozen different distros I've tried in the past few months. From an end-user's perspective, Linux just comes across as too fragmented; plenty of good, but less than great options, leading to significant diverges in system behavior. I realize that Linux wouldn't exist if it were not for the freedom to do this, but at the same time, it's the main thing that causes Linux to lag behind.

I'd find it hilarious if in their Linux expeditions Microsoft winds up making a 'standard desktop' and then ports over Office...
 
But that's still just theory?

That bolt on modules affect performance, etc.- because in theory, they must, but in practice, is the difference measurable?

Beyond that, I have no issue with Nvidia protecting their IP through closed-source drivers so long as said drivers interoperate well and do not hinder performance, and from what I've seen, they work great.

I also have no issue with AMD doing what they can, in this case garnering sentimental support by opening up their code. I wish them luck, I just don't fawn over them through some alignment of ideologies. I'm glad Intel is more or less following suit, but again, they also really have to. I expect my next laptop upgrade to be another Intel machine, but with a 120Hz VRR display. AMD if they can get all of their ducks in a row the way Intel has.

Well unless Nvidia is going to provide an open source version to the kernel... there is no way to test things in their case. However in general most hardware manufactures drivers start as bolt on in the testing phases before being passed to the kernel team for inclusion in mainline. Yes there are performance advantages. (not to mention not forcing end users or distro maintainers into having to build custom kernels all the time)

Just to be clear though Intel has always been open source friendly... long before AMD was. Intel attempting discrete gpus again is new is all. :) Anyone with current integrated intel cards don't need to worry about nothing... everything official is in the kernel. That isn't expected to change. If you run an Intel GPU if you build a kernel from source your good to go. Pretty much every Linux distro runs silky smooth on 100% intel hardware out of the box with stock kernels.

I'd argue the complete opposite- Apple being the sole curator of their 'experience' seems to provide the most coherent one, and Windows 10 is still far more consistent and responsive than any of the dozen different distros I've tried in the past few months. From an end-user's perspective, Linux just comes across as too fragmented; plenty of good, but less than great options, leading to significant diverges in system behavior. I realize that Linux wouldn't exist if it were not for the freedom to do this, but at the same time, it's the main thing that causes Linux to lag behind.

I'd find it hilarious if in their Linux expeditions Microsoft winds up making a 'standard desktop' and then ports over Office...

To me "User experience" doesn't mean joe average. (only anyway) Apple has done a good job creating a UI for joe average yes. To me user experience with Linux means every company in the world that wants to use it and build a system on it can very easily. I think we are talking two different languages on that one. For the record Linux is NOT lagging behind. Its the #1 in use kernel (Linux is a kernel folks) in the world and #2 isn't even close. Love or hate Androids UI... average type end users don't see Linux anything. As for desktop Linux... ya there are lots of options that isn't a bad thing. (talking about Operating systems that aren't shipping with hardware are not really all that relavant imo) No one is saying ChromeOS is hard to use and fragmented. The kernel is easy to use... and saves companies like Google billions in R&D.

On MS yes I have been saying for a few years now... even before the Linux subsystem stuff. Its only a matter of time. MS will at some point use the Linux kernel as well. They can't out spend every other company in the world. Which is what they are currently trying to do. Maintaining windows is a loosing proposition. They have already mostly lost the server market. Their own big iron running cloud is powered by Linux. They lost the mobile market a few years back and it doesn't appear (wisely) that they are going to attempt to make another push there. At some point they will realize trying to outspend Google/Amazon/Facebook/Samsung/IBM/AMD/Intel + every other Linux using company in the world is silly. Develop a Microsoft closed source UI... conform to the FSH, and run the kernel on the back end. 99.9% of their windows customers would never even notice the difference, but it would slash the costs of maintaining windows significantly. (and make it 100x better) It would be no different then Apple using darwin (next step / unix) at its core... end users have no idea they are using a 100% postix compliant Unix system.
 
Last edited:
To me "User experience" doesn't mean joe average. (only anyway) Apple has done a good job creating a UI for joe average yes. To me user experience with Linux means every company in the world that wants to use it and build a system on it can very easily. I think we are talking two different languages on that one. For the record Linux is NOT lagging behind. Its the #1 in use kernel (Linux is a kernel folks) in the world and #2 isn't even close.

Agreeing largely with the other three points, but yeah, we're speaking two different languages. User experience is just that- Linux is a bit of a shitshow there, whereas Windows and OSX are both pretty well put together.

Here's one example: an elevated file manager in the GUI. First you have to figure out what the installed distro is using, then what the process is actually called... yeah. Option to just run an elevated file manager? Doesn't exist. Option to do that when you do something that requires it, or hell, just a prompt to tell you that you need to do that?

Just random inconsistent stuff; it's not a big deal if you're only dealing with the same handful of applications on an already built system, like installing Mint and using what's there and needing nothing more, but there's a pretty large gap between the level of troubleshooting a beginner can accomplish and what a proficient user can do without excessive web searches.

I'm still stumped on the Wine causing audio drop-out thing on my ultrabook...
 
MS will at some point use the Linux kernel as well.

Addressing this specifically: I do think it will take the better part of a decade for this to happen. Not really because Microsoft chooses to go this direction for their desktop/server products, but because the computing paradigm itself will shift away from the desktop.

That said, Windows 10 is at a pretty good spot now. Server 2019 is too, for those things that it excels at, so I don't see Microsoft transitioning away from Windows kernels inside of two or three decades really. How they package it might change- but it works very well.

I also don't see any issue with them transitioning to using the Linux kernel (and when we say 'Linux', we do mean GNU/Linux...), which they may decide to do experimentally regardless.

However, with the transition toward 'many core' computing, the idea of containerization where the host kernel simply doesn't matter may just push us past that. Shared/common frameworks also help, to the point where applications can easily be built for multiple platform (ARM, x86, Power...RISC-V?) and kernel backend combinations, if they're not just outright written for web use in the first place.

At which point the Windows kernel may exist so long as Microsoft chooses.
 
At which point the Windows kernel may exist so long as Microsoft chooses.
But it will never be able to come close to the Linux Kernel and money is not the problem, Microsoft is ...
There will be a point where Linux with all of the features which we now enjoy on Windows will surpass Windows because of all of the things that been biting Windows in the ass (MS programs with kernel access, telemetry access that ignores security features) .

Freesync is just the next nail in the coffin..
 
But it will never be able to come close to the Linux Kernel and money is not the problem, Microsoft is ...
There will be a point where Linux with all of the features which we now enjoy on Windows will surpass Windows because of all of the things that been biting Windows in the ass (MS programs with kernel access, telemetry access that ignores security features) .

Freesync is just the next nail in the coffin..

Freesync finally shows up in Linux years late but it is the next nail in the coffin? LOL :D
 
Freesync finally shows up in Linux years late but it is the next nail in the coffin? LOL :D
Don't forget that the Linux kernel in some form is almost in every device :) it is so small it is so versatile....
 
There will be a point where Linux with all of the features which we now enjoy on Windows will surpass Windows because of all of the things that been biting Windows in the ass (MS programs with kernel access, telemetry access that ignores security features) .

This has become a business model in a sense for Microsoft, but none of this is inherent, and it can also go away, so we cannot say that there 'will' be a point, but that there 'could' be a point. And there could be!

I'm just pointing out that it is far from certain; Microsoft could easily switch business models and retain their desktop market, or even grow it, if desktops continue to exist.
 
I means you can use 100% free open drivers and use freesync. AMDs open source stack has had a lot of work poured into it. To the point where its faster in many cases then the closed source AMD kernel module option. In comparison Nvidia as an example provides close to zero support to their free in kernel developers... they provide just enough support that Linux will boot allowing people to bolt on the Nvidia kernel module.

On the kernel stuff... IMO everyone should be on manjaro.

Having said that... all the major distros have fairly easy way to run the latest verisons of the kernel. Yes their "stable" versions may ship with an older LTS... but anyone that is smart enough to want to turn free sync on (as you point out with free drivers) can handle switching to a newer kernel line.

Don't forget the fact that AMD do this because they can't be arsed creating decent drivers for Linux, so they hand the community a bunch of documentation and state "here, do it yourselves, we really don't care enough for you to put in the hard yards."
 
And almost no desktops- why bother with the mischaracterization?
It only shows that what MS does (their kernel) is irrelevant because it does not scale. If you know how things work right, more code = more change of bugs.
And that is the only thing that counts.
 
It only shows that what MS does (their kernel) is irrelevant because it does not scale. If you know how things work right, more code = more change of bugs.
And that is the only thing that counts.

It scales just fine?

It just isn't designed by Microsoft- currently- for the extremely limited applications involved in big iron. There's no evidence that it cannot be; Server 2016 and newer are highly performant. The biggest differences seem to lie more in terms of business model and development approach rather than real technological limitations, and if there are real technological limitations, there's no evidence that Microsoft couldn't overcome them if they so chose.
 
Don't forget the fact that AMD do this because they can't be arsed creating decent drivers for Linux, so they hand the community a bunch of documentation and state "here, do it yourselves, we really don't care enough for you to put in the hard yards."

That really depends on your idea of decent Linux dirvers. Their Linux drivers are very stable and solid. They just aren't gaming drivers.

This is why OPEN is best for gamers. The money will never again be in games for GPU manufacturers. Even Nvidia. I'm not saying gaming is a zero market... just that the AI and neural n markets are already huge and only growing. AMD takes care of their commercial linux users... they don't care if X or Y gaming title is 5-10% faster with a new driver. They care that things are rock stable and when they use their GPU for a render assist it isn't Fing up pixels.

AMD could benefit from 100% open source drivers for windows as well... there just isn't the developer base willing to do their work for them in that pool.

Nvidia may be the big gaming dog today... but that can change. If it does and Nvidia is seeing 80%+ of their profit from non game things... they won't care about game drivers anymore either. Open source is the best option for consumers. One Graphics stack that works for everyone... all the GPU manufacturers should HAVE to do is supply low level access to the kernel. Everything higher then that in the stack should be standards based open source. It sounds like pie in the sky... but AMD and Intel prove its possible. If Nvidia did the same.. MESA could be the universal stack targeted by developers and gamers would win.
 
It scales just fine?

It just isn't designed by Microsoft- currently- for the extremely limited applications involved in big iron. There's no evidence that it cannot be; Server 2016 and newer are highly performant. The biggest differences seem to lie more in terms of business model and development approach rather than real technological limitations, and if there are real technological limitations, there's no evidence that Microsoft couldn't overcome them if they so chose.

Well actually there seems to be plenty of evidence. MS has tried to go after that market in the past and fell pretty flat. They can't even get a proper cloud/big iron file system to work properly.

I would say if MS was capable of going after the market they would have. They like profit.

Understand I'm not knocking MSs kernel really hard. Just that long term there is almost no way they can keep up. The main advantage for years has been an army of cheap hardware manufacturers building drivers aimed at their micro kernel and ignoring most other consumer devices. Thing is that isn't really the case any more with Apples rebuild... and more importantly Android/ChromeOS. It means even manufacturers making the analogus of 2018s software modems in general are supplying drivers to the kernel for android support.

I don't think MS switch to using the LInux kernel is decades out as some may think. I am also not suggesting their "windows" product will be an Ubuntu like distro. Its not going to use GNU... just the kernel. Perhaps the FHS so they don't have to deal with the mess of trying to do their own thing there. Most users won't even notice the difference if they do switch. It will look the same act the same... perhaps have a different directory structure.
 
MS has tried to go after that market in the past and fell pretty flat.

They're still getting there- going to a full terminal-based install, for one, and supporting all kinds of virtualization stuff. As for a cloud/big iron file system... I've tried ReFS, not at scale, but it works great for what it is designed. ZFS is getting at the same thing, as is BTRFS- and not a one of them are perfect. I run ZFS right now, only because I was interested in it and FreeNAS (being free) brought many of the features I was looking for. I'm also interested in learning...

But really, I see Microsoft as moving toward having a largely 1:1 solution with respect to what GNU/Linux currently offers. Obviously they target sales, which the Linux community doesn't directly, so I don't expect them to see parity in either direction, but I do expect it to be close. ReFS, for example, is in some ways more flexible than ZFS in FreeBSD or in Linux, or BTRFS in Linux, and I wonder which will reach full functionality first.

[I don't expect Microsoft to go after big iron- it's such a small absolute market, with each implementation being so customized, that I doubt they'd see the utility- but there's really no reason they couldn't should they find an application]
 
That really depends on your idea of decent Linux dirvers. Their Linux drivers are very stable and solid. They just aren't gaming drivers.

Beyond Wayland support, there's no problems with Nvidia drivers beyond what the open source zealots like to claim as the drivers are binary blobs which doesn't sit well with the zealots. In relation to Wayland, who cares? Wayland is still in the early days of Pulse Audio development as far as I'm concerned.

When are the open source drivers going to get a decent control panel?
 
Beyond Wayland support, there's no problems with Nvidia drivers beyond what the open source zealots like to claim as the drivers are binary blobs which doesn't sit well with the zealots. In relation to Wayland, who cares? Wayland is still in the early days of Pulse Audio development as far as I'm concerned.

When are the open source drivers going to get a decent control panel?
They (well, AMD) are working on implementing the features needed to have something like catalyst control center on Linux, but I don't expect it'll be soon. After they implement the features, they still have to make a ui. That post was from a year ago, I don't expect a whole lot has happened since then, but maybe there was some progress.
 
That really depends on your idea of decent Linux dirvers. Their Linux drivers are very stable and solid. They just aren't gaming drivers.

I've been fine with the performance of the Open Source drivers and my RX 580. Granted, I'm also getting old and spend a lot more time playing strategy games than shooters.

The only things I feel I'm missing from Windows are Eyefinity and Freesync.
 
  • Like
Reactions: ChadD
like this
And a decent control panel to implement them with... ;)

One of the reasons I like Ubuntu is that most of the things I would use Catalyst for and used to be a pain on Linux (TV adjustments, switching between 4k 60hz/1080p 120 hz) are integrated right into settings under Devices->Displays. If they could add Freesync and include Eyefinity as a supported display mode I would have no need for the CCC.
 
Back
Top