AMD Still Struggling With Linux Gaming 100
An anonymous reader writes: AMD's Linux gaming performance has been embarrassingly bad, and it doesn't look like there's any quick remedy. Virtual Programming just released Dirt: Showdown for Linux, and it's the latest example of AMD's Linux driver issues: AMD's GPU results are still far behind NVIDIA's, with even the Radeon R9 Fury running slower than NVIDIA's aging GTX 680 and GTX 760. If a racing game doesn't interest you, Feral Interactive confirmed they are releasing Company of Heroes 2 for Linux next week, but only NVIDIA and Intel graphics are supported.
Come AMD.... (Score:4, Insightful)
Get it together, competition in the market place is good for us all!
Re:Come AMD.... (Score:4, Insightful)
Re:Come AMD.... (Score:4, Insightful)
Hence why they're putting so much effort into open-sourcing as much of their drivers as possible, while working with the community to make what remains a binary blob more secure. It takes time to fix your mistakes and get your shit together, but that doesn't mean they're doing nothing about it. As a bonus, the same effort may also potentially be helping to fix their shitty Windows drivers.
Re: (Score:2)
ATi is part of AMD now.
Re: (Score:2)
I partially agree with you. AMD/ATI graphics can run as good as nVidia under certain circumstances. Generally not with OpenGL and not with non-Windows drivers. ATI has never been good with providing Linux drivers, and what they did in Windows was mostly for DirectX.
nVidia and Intel have both done a bunch to ensure that AMD and ATI are playing on an unfair field. I still boycott Tom's Hardware because they were so extremely dishonest with benchmarking.
Re: (Score:1)
That being said, nVidia only has closed source Linux drivers which may not be as bad as AMD's, but they are not great either.
Re: (Score:2)
AMD has much more to gain from good open source drivers than Intel does, and currently Intel is way ahead of any graphics hardware maker.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
Straight Outta Your Ass
This is the same desktop market that Microsoft now claims is dead? The same desktop market that OEM's ship with piece of shit Windows on? The same desktop market that people with Mac's left in droves because Windows sucks?
News flash. You can add Linux to any Windows desktop for free. distrowatch.com It's not shipped on your PC because Microsoft shitware likes to pretend their market share is due to superior code. (pspspsss it's there because of dickmove money deals) You can als
Re: (Score:2)
Re: (Score:2)
they are fixing it... they new driver policy is to have a generic driver in the kernel. then you plug mesa or catalyst on top of it.
The idea is that mesa is getting better and better and support more cards, mesa support up to opengl 4.2 is expected to be release in September. with more features ready, performance is also getting better. For possible more performance, features or too new hardware for mesa, you can plug catalyst and use the closed source driver. With this, even the closed source linux driver
Re: (Score:1)
There's ARMs for competition and that's who intel changing their business strategy because of them
Open source the problems! (Score:2, Insightful)
Open source the problems!
Many eyes will fix it.
Re: (Score:2)
However, even if they can, it may be a problem of design. If AMD's drivers were built with a paradigm that only really works on Windows, just being open source isn't going to be of much value. The driver would still have to be redesigned.
Re: (Score:2)
Err thats why there is already an open source driver?
Open sourcing more of the specs (some specs are already released) would allow that driver to improve in leaps and bounds.
Re: (Score:2)
Re: (Score:2)
Insightful, Troll, Interesting, what you lack is a Funny moderation despite that's what it is. AMD has open sourced. While there's bits and pieces missing for specialized functions you have all the low level shader instructions to implement high performance gaming. Turns out it's tough work and not a whole lot of people who actually understand how to make efficient use of a modern GPU's resources. And those who do generally are or have been employed by AMD/nVidia/Intel and has to deal with a lot of thorny I
Linux has been embarrassingly bad (Score:1)
And it doesn't look like there's any quick remedy.
FTFY
Re: (Score:2)
Busy on DX12 and Vulkan (Score:2, Insightful)
I have no inside info but it seems obvious that for a while now (pretty much most of this year) the focus has been on DX12 and Vulkan. I think Graham Sellers is on the GL team and he is definitely knee-deep in Vulkan.
The New Shiny (Score:3)
Yes! The NEW SHINY will SOLVE ALL OUR PROBLEMS!
Filter error: Don't use so many caps. It's like YELLING.
Re: (Score:1)
Actually, it has. But how would you know it?
Re: (Score:2)
Yes! The NEW SHINY will SOLVE ALL OUR PROBLEMS!
Well if you actually understand it then you understand why GP's comment is pretty valid and that taking a cynical view just because you don't know any better is pretty silly. Much of the problem of driver performance and complexity is the the current APIs are not an accurate abstraction of modern hardware. So the driver contains a lot of code to convert from what the developer writes for - the hardware view they get from the API - to what the actual hardware is. A lot of that code is application-specific to
Re: (Score:2)
When was the last time the NEW SHINY actually did SOLVE ALL OUR PROBLEMS?
It's always going to. So far, in my experience, it never has. Tossing out an established code base and starting over is almost always a bad idea.
Re: (Score:2)
When was the last time the NEW SHINY actually did SOLVE ALL OUR PROBLEMS?
I'm sorry I'm not across every single new invention but in the case of this domain then certainly, as I already mentioned, the switch from fixed function to programmable pipelines has been an enormous improvement but obviously that falls outside "your experience". So why do you believe that replacing an API design that presents a representation of antiquated hardware coupled with a heavy translation layer to convert this to the real hardware, with an API that much more closely replicates the underlying hard
Re: (Score:1)
Your post made me ponder so I ended up scrolling back up to find it. Over the years I have figure out that, for me, it seems that the more bleeding edge your hardware is the more likely you are to be cut. This is, of course, part of the adage. Anyhow, this seems particularly true in the graphics arena. I am not sure why that is.
Re: (Score:1)
Re: (Score:3)
AMD shoul just release the COMPLETE code already. Then maybe the community can begin fixing all these problems. And don't get me started about how they released the code. It's not all there and is largely a joke. Release some code and wrapping it around a proprietary component is hardly really "open source".
Not really sure what you're grasping at here. The Catalyst driver is not and will not ever be released, due to a number of reasons ranging from trade secrets to IP issues to DRM issues to whatnot. The specs have been released and an open source driver based on it, thers' no "proprietary component" it revolves around. Unless you talk about the firmware which is pretty much like every other hardware company, they load a blob that sets up the hardware correctly. Just about any modern hardware has this, it's ju
Re: (Score:2)
But AMD gets much of the same flak despite making much more of an effort.
We understand that there are legal complications regarding releasing the information, but if they can't do that and they can't make a decent driver themselves, which it appears they can not, then yes they're going to get plenty of flak. Why wouldn't you expect competence?
Re: (Score:2)
We understand that there are legal complications regarding releasing the information, but if they can't do that and they can't make a decent driver themselves, which it appears they can not, then yes they're going to get plenty of flak. Why wouldn't you expect competence?
Why do you need their driver - that you admitted isn't decent - when you have the specification to work to?
Re: (Score:2)
Why do you need their driver - that you admitted isn't decent - when you have the specification to work to?
Because I'm not much of a programmer, and have no driver programming experience, and graphics drivers are serious "there be dragons here" territory at the best of times, let alone with potentially incomplete documentation.
I don't have the allergy to binary drivers that some have. I'm pretty happy with the nVidia drivers. They're not perfect, but what is?
Re: (Score:2)
Re: (Score:2)
because there isn't a full specification to work to. did you even read what you were replying to? they don't release _enough_ information to make a decent open source driver and their closed source driver sucks ass. maybe they don't even know themselves and their drivers are a mishmash of code from subcons. maybe that's why they were trying to do mantle.
Re: (Score:2)
because there isn't a full specification to work to. did you even read what you were replying to?
Of course I did, the title and the comments in the thread.
they don't release _enough_ information to make a decent open source driver
What else do you need?
Re: (Score:1)
Suddenly (Score:4, Interesting)
...2 games appear not to be supporting AMD graphics processors. That's strange, nearly 1.5 thousand from steam do not have a problem at all.
OTOH, intel's iris' drivers are a joke yet noone bats an eye.
Re: Suddenly (Score:2, Insightful)
Intel has much, much better drivers than AMD. Intel gpu hardware isn't nearly as good, unfortunately.
AMD = Awful Motherfucking Drivers
Re: (Score:2)
Intel's integrated GPUs are horribly, horribly slow if you're expecting performance anywhere even vaguely near what the big kids in graphics are doing. Any time intel tries to do graphics, they throw CPUs at it, and it ends up costing a mint. Or, they buy someone else's graphics, and they turn out to be crap.
I haven't touched a modern ATI video card in ages, so maybe they've gotten better at writing the software part, but I know nVidia is still capable, so I have no reason to risk finding out again.
Re: (Score:1)
Intel in general had better drivers than either AMD or nVidia. Unfortunately it's their hardware that's often lacking or not competitive.
That said, Intel also has a few GPU's that lack proper accelerated drivers for Linux and/or gave sucktastic performance.
Personally, I'm looking forward to when the "mobile" graphics vendors start to become more and more competetive with the big boys. Since those generally have Android drivers a port back up to linux-general shouldn't be too hard.
Come on guys, have some understanding (Score:1)
1) They can't release complete sources, because of, cough, licenses
2) Linux gaming is a tiny piece of the market, yet AMD is financially troubled, do you really want them to spend much resources on less tan 1% of the (gaming) market?
Re: (Score:2)
Why does having to embrace Linux == embrace openness?
Linux != openness.
Well, you're half right. It's ironic you end with a jab at RMS, because the GPL is part of what defines Linux. Remember, PC-BSD was a thing. Minix was a thing. By the time people really started to jump on the Linux bandwagon, there were alternatives. But what made Linux different was the license. People had an expectation that their work would not be hidden away from them. People who would not have otherwise contributed did so; they have since said so.
Beyond that, though, AMD has never made a serious effort
Re: (Score:2)
Linux community itself, for the most part, embraces openness, but the concept itself isn't for that.
Actually that's where you're dead wrong. In 1985 RMS outlined the goals of his free software, and the freedom to study what it does requires openness. Now various companies have found loopholes, but the fundamental goals remain the same: You should be able to run, study, improve and redistribute the source code. And yes, study implies being in the source form most suitable for viewing, not assembler instructions or whatnot.
Re: (Score:3)
It's not all that uncommon actually. I work for a CPU vendor and we incorporate intellectual property from other sources into our chips, typically Synopsis for things like USB, AHCI, PLLs and other areas. Fortunately we are able to document everything and we fully support Linux and release the source to our bootloader and SDK. I maintain the bootloaders for many different boards and I've had to re-write a number of phy chip SDKs (usually 10G) since they were not compatible with the GPL or our SDK's license
Re: (Score:1)
You make too much sense Kartu
Re: (Score:1)
Never mind the gaming (Score:2)
If Microsoft are not bribing people to put out this crap, it is even more shocking than if they are!
Re: (Score:2)
Hey, at least you get drivers. They stopped supporting the integrated graphics in my AMD machine a couple of years ago.
Re: (Score:2)
Hey, at least you get drivers. They stopped supporting the integrated graphics in my AMD machine a couple of years ago.
Don't feel bad, I've got an AMD-based netbook that was never supported and still isn't. It only works kind-of OK with Windows, though...
Re: (Score:2)
Re: (Score:1)
Crap Article. (Score:4, Informative)
It's an eON wrapped game with nVidia support scripts. So obviously nVidia cards will have better performance.
Why don't they test with some native games. Sure nVidia will probably do better. But AMDs will do good enough.
crap article (Score:1)