Slashdot Mirror
Open Radeon 3D Driver Runs At 60~70% of Proprietary Driver Speed
An anonymous reader writes "AMD's Radeon HD 6000 series open-source Gallium3D driver for Linux is now working and running at 60~70% (in some cases, 80%) of the speed of the official proprietary 'Catalyst' driver. This is a big speed improvement in Mesa/Gallium3D compared to the times when the performance was crippling or even just a few years ago when AMD didn't support open-source drivers. When will NVIDIA change ways?"
Given that ATI's proprietary driver also runs at 100% of its own speed, there are apparently motives that apply in spite of that...
because it sucks.
Many reasons..
The binary driver cannot be redistributed with the linux distros..
The binary driver may drop support for older hardware at any point, and the older versions which still support your hardware are unlikely support current kernels or X11 versions.
You cannot fix a binary blob driver yourself, you are beholden to the vendor to do so.
Also that "100%" is relative to the binary driver itself, its possible that given time the open driver will surpass it.
Out of interest, does the open driver support OpenCL yet?
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
Remember, it's Phoronix.
Yet Another Tech Blog
(but so much more, including game and movie reviews)
http://yanteb.peasantoid.org
As best I can tell, the parent poster was replying to TFS's question of "When will Nvidia change its ways?" with "Why change when their[Nvidia's] driver runs at 100% of it's speed?".
My response was that, ATI had apparently found some reasons to change, ie. start supporting an OSS driver effort, despite the fact that their[ATI's] proprietary driver, just as with Nvidia's, had always run at 100% of its own speed.
Presumably, something other than sheer performance considerations is behind the fact that Nvidia 'supports' OSS drivers in the sense that their cards function well enough in VGA mode that you can make it to Nvidia's website to download the proprietary ones, while ATI(or AMD) seem to be making real headway in encouraging OSS drivers of real-world utility.
Because proprietary drivers traditionally have minor bugs and annoyances which are getting fixed like never. Not everybody is craving for the top fps on the new games - many want speedy 2D and video without glitches. I'm not sure that OSS drivers would be better in the respect, yet IMO chances are better with two alternative drivers available.
Also, OSS drivers for both nVidia and ATI would likely exchange patches or probably reuse many common features, making them more compatible to each other, thus reducing number of surprises when something works on ATI driver but not on nVidia's.
All hope abandon ye who enter here.
I use Linux on a Powerbook with Radeon graphics. For some odd reason, AMD does not provide binary drivers for this platform, but the open driver works great.
On my other machines with x86-64, I use the binary Radeon drivers as it is the only way to get full OpenCL. Even there, it sometimes happens that I need to use an older kernel or disable some kernel features, as the binary driver does not play well with the pace of Linux development.
Escher was the first MC and Giger invented the HR department.
I have CAD at home on Linux (Draftsight for 2D and Varicad (It's Linux native!!) for 3D), and there's no substitute for the Catalyst driver. The free drivers don't cut it. They may cut it for generic desktop stuff like playing video and spinning desktop cubes, but somehow combining the free driver and any CAD package gets you a very slow experience.
Until performance really does reach 80 percent, I'm gonna have to stick with the proprietary one. And since this is only for the 6000 series and not the 4000 series (my card), I'm just gonna have to forget about it until I get new hardware.
Hands up if you've ever had to call the ATI BBS in Peterborough, ON back in the day to get the driver of the week for Mach32 on any system.
By the way, if you want free 2D Cad for Linux, get your ass over to Dassault Systems and download Draftsight.
--
BMO
It's not like you're going to be doing alot of gameing on those machines. So honestly. who gives a shit if it's not at 100% of its speed...
Oh right. delusional geeks. carry on.
No not yet openGL is the current priority, there has been a start made but it seems thy want to get the drivers working for graphics first. on the other hand they have just reached openGL 3 my understanding is that priorities are going to start to shift now, video decode acceleration is currently being worked on so I guess thy may start work soon.
The nvidia-glx package has been broken for a week or two in Debian wheezy, and the nouveau driver makes my GPU fan spin as if I was trying to calculate a trillion decimals for pi. Result: I am booting Windows XP right now. Guess who's going to be buying ATI next time he replaces his computer?
I'd use it.
The single biggest source of problems I've had with Linux over the years has been hardware without open-source drivers. I'd go so far as to say it's the only source of unsolvable problems I've had with Linux.
If you're not a serious gamer, you don't need a card to work at 100% of its potential - 99% of the time all I need mine to do is use about 1% of its power to render a desktop. If the driver's reliable and open-source, why would I care that a different driver would give me a slightly-better FPS on the very rare occasions that it's working flat-out?
So.. it has come to this
Uh the proprietary drivers are still free.
Meanwhile I have a Gateway subnotebook with AMD processor and graphics (Athlon 64 L110, R690M with X12xx graphics) and it only runs Vista. On Windows 7 resume doesn't work. On Linux I get complete graphics corruption even with RenderAccel disabled. So for some of us, neither fglrx nor ati works. Which is why some of us have finally learned our lesson, and will never buy anything more complex than a cellphone with ATI graphics again.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Being open means that these drivers won't simply go away once the product line is deprecated in favour of the newest and coolest graphics card, and that it will be able to receive improvements and bug fixes essentially until the last working piece of hardware dies off. Being open also means that it will be able to provide support for this Radeon graphics cards in other platforms besides the officially sanctioned ones, such as Windows and Linux. Being open also provides a way to provide competition for the people AMD employs to develop their official graphics card drivers, because if an open driver developed by amateurs on their spare time happens to be nearly as good or even better then they may as well be out of a job, and they can't have that. Being open also means that, if the open drivers mature enough so that they are comparable to AMD's official offering, then it will be in AMD's best interests to get directly involved in the development of these open drivers and even abandon their proprietary offering in favour of this project.
And, obviously, if these open drivers represent a business success story to AMD then you can bet that this will spread out to other companies, and everyone who used windows and had to deal with hardware with support problems certainly knows what a PitA it is to be tied to proprietary drivers which are crap.
Slashdot, fix your code or at least hire someone who is competent at it to do it for you.
Dammit I do not know - but it is not going to be soon; how could they opensource it and prevent disclosing secrets to their competitors at the same time?
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
ATI proprietary driver for linux, at least the Radeon 5x mobile whatever that is in laptops is utter FAIL.
.png objects, a chat window that can roll dice... with the ATI Catalyst driver , once the map got bigger, it didn't bother to redraw screen in places, leading to new messages in chat being hard to notice without constantly messing with the window and corruption of graphics elsewhere...
Yeah, it has nice fast performance of glxgears, but that's about it. All the games i have played with an NVIDIA card under WINE, like World of Tanks, and Starcraft II... they become either unplayable due to texture corruption, or outright crash when executing.
The second it was a hell to install simply due to failure to render the fancy installation graphics.
Well, that'd have been somewhat bearable, only non-native apps... but , MapTool , the program i use for DnD sessions... nothing fancy , just a map with a bunch of simple
The opensource one for this series was slow as hell, (tenth of performance) and somewhat buggy too, but at least the above program worked.
You can do framebuffer output with the proprietary drivers installed just fine. The issue with lack of KMS support is there's always a bit of graphical glitching as you transition between the framebuffer, and the rendered X display. KMS gives you a smooth transition from one to the other.
If you're going to forgo a significant amount of graphical performance just to remove some superficial glitching that you only ever see on boot, then you are a vain anonymous coward.
I seriously doubt Nvidia is actually going to change their ways. Nvidia claims that a significant amount of their graphics technology IS the driver and that opening it up would expose too much of their IP. AMD doesn't really seem to hold this view so I'm guessing their secret sauce is more on the hardware end. In my opinion, AMD probably even has the better hardware, as seen by how it scales more linearly up to higher resolutions. Nvidia manages to come up with enough quirky driver optimization to stay competitive.
Fear is the mind killer.
The advantages are more nuanced than people think. Sure, the proprietary drivers have better raw performance, but the FOSS drivers are (as you said) much more stable because they're generally updated in step with kernel and Xorg updates, which means that their compatibility is much better. Both the nvidia and catalyst drivers, on the other hand, tend to get updated more infrequently and more in a pattern of being fixed when obvious bugs crop up, not being updated to avoid bugs even existing.
Pretty much the only worthwhile argument you can make for using the proprietary drivers is if you commonly use a feature that the FOSS drivers don't support for your card version yet, which can be common with newer cards.
Yet Another Tech Blog
(but so much more, including game and movie reviews)
http://yanteb.peasantoid.org
You're completely ignoring the big fact that the Open Source Radeon driver also runs at 100% of its own speed. In fact, I'm willing to bet good money that ANY driver can run at 100% of its own speed.
Nope. It's a proven fact that adding a HOSTS file will improve the speed of any driver to the point that it runs at 150% of the speed of itself. The fact that this will cause a rip in the space/time continuum, making the universe implode is irrelevant, because the HOSTS file also creates 100% security from thin air, so the driver will continue to exist in perpetuity.
HOSTS files FTW! Is there anything they can't do?!
"City hall" in German is "Rathaus" Kinda explains a few things......
somehow combining the free driver and any CAD package gets you a very slow experience
Not quite true for me: Maya and Blender work fine. In fact, Maya is noticeably faster that it is on Windows 7 on the same machine, and I have taken no special steps to make it faster on Linux. I am using the laziest Linux distribution there is (Wubi Ubuntu, click click click until you have a working OS)
Minecraft is faster and more solid too (not exactly a CAD, but it uses GPUs if I am not mistaken)
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
...it's only advantage is being Open?
I can see how many people may not see a great cost/benefits ratio there...
The main advantage of being open is long term support. Graphics card drivers are quickly abandoned by AMD once they are a few years old. So their newer drivers don't support old cards, and older proprietary drivers don't support new kernels. So your only solution when using an old card (pre HD series) with a new operating system is to use the open source driver. The problem is not limited to Linux. On Windows, AMD issues "legacy" drivers for older cards but they are not thoroughly tested. So while they fix compatibility with some software, they break it with others that were working great with the old drivers... Worse, there's no support for them. On my Linux distro, while using open source Radeon drivers, performance keeps improving with each new version.
You can pass a vga= argument to the kernel on boot to allow modes other than 80x24. See this table for possibly modes.
--srj/mmv
Careful, you may invoke APK
(1.21 gigawatts) / (88 miles per hour) = 30 757 874 newtons
You're right, the current open drivers do not cut it, by far. But the ones mentioned in the article are not included yet in any major distro, do look for them in the next Ubuntu. As for your 4000 series card, it's probably time to upgrade to 6000 ... I just did for playing wine games and the difference between the two is pretty incredible.
For games, I will stick with the closed driver until the open one is just as good or better. AMD has promised to make the open driver have the same development cycle as the closed one for the 8000 series cards. Since they promised some years back to help the OSS guys build a driver, and have delivered on that promise, I'm optimistic this will happen.
At that point we should see the performance of the open driver be just as good as the closed one, with the advantage of more testing : better stability and security. It could even surpass the performance of the closed driver.
Good move on AMD's part, and I will be continuing to support their video hardware. Now if only they would release a CPU I actually want ...
Being open means that these drivers won't simply go away once the product line is deprecated in favour of the newest and coolest graphics card, and that it will be able to receive improvements and bug fixes essentially until the last working piece of hardware dies off.
What it actually means is that bug fixes and improvements will be possible as long as someone is willing to work on them. Even if the hardware is around, just like AMD, someone has to be willing to do the coding
Interestingly, nVidia is actually pretty good at fixing bugs.
GNOME3 had a nasty corrupt-on-resume problem with the nVidia driver, and since a) laptops are slept and resumed often, b) nouveau has no power management to speak of, which is kinda important in a laptop, and c) the GNOME devs had no intention of fixing the problem anytime soon, it was nice that d) nVidia fixed it in a month. They're pretty good with other bugs, too.
The nice thing is that, with GNOME3 and nVidia, I have the first instance of tear-free video playback on a Linux desktopin, wel, ever*.
I don't know it AMD/ATI better now, but Catalyst used to be brutal for bug fixes. I think they're better. I also don't mean to impugn Nouveau as they've done great work with what they've had, but I do value battery life and not cooking my thighs.
* without turning off compositing and partying like it's 1999.
--srj/mmv
Maya and Blender both are not CAD.
And Minecraft certainly isn't.
There's something about CAD that drives the free drivers over the edge.
Like I said, in some situations, the free driver is certainly enough.
I consider the driver as part of the card. I can't modify the card itself, and it doesn't bother me that I can't go mess around in the proprietary driver changing things around. In an ideal world, self help for fixing proprietary drivers would happen, but both ATI and Nvidia think that keeping features locked away is a competitive advantage because it keeps "the other guy" from copying hardware or such.
--
BMO
And yet the open driver supports them anyway.
The more widespread open source becomes, the more practical alternative architectures become.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
Maya and Blender both are not CAD
CAD = Computer Aided Design, right? Am I missing something here?
keeping features locked away is a competitive advantage because it keeps "the other guy" from copying hardware or such.
I could not agree more (see my post further below)
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
Funny, when I use the free drivers my screen flickers.
Give me Classic Slashdot or give me death!
Don't burst his bubble, that's not nice.
I wouldn't say that's fair, the Xbox 360's graphics work fine! (Which cellphones are starting to come towards in graphics complexity..)
>CAD = Computer Aided Design, right? Am I missing something here?
Yes.
--
BMO
If you take a look at the xorg's mailing list you will find out that drivers for ancient relatively rare graphics cards such as the Matrox 400 line are still being developed and maintained. So, I don't believe that no one will be interested in getting involved in this sort of project.
Slashdot, fix your code or at least hire someone who is competent at it to do it for you.
Because proprietary drivers will taint the kernel, and you may lose support because of that. Because one can't fix bugs in a proprietary driver, and no company will fix them fast enough or on old drivers. Because you want to add something to it. Because you want the driver that was compiled toghether with Xorg and your kernel, so anything wrong appears before distribution. Because you want the driver that (again) was compiled toghether with Xorg and the kernel, so that you'll be sure there will be no delay between the compilations and your driver will be fit to the version of the kernel and Xorg you are using.
Well, I've never had any reason out of the above. Other people experience may vary.
Rethinking email
Careful, you may invoke APK
You mean foaming saliva boy? That's kinda the point. :)
"City hall" in German is "Rathaus" Kinda explains a few things......
Somehow, that adage almost never works well at the real world.
Did you get some Greece bounds a couple of years ago?
Rethinking email
Except.. These drivers have been around for years, and don't. So you wait for what, 3 years to get a driver that performs just over half as good.
Hey, here's a bug report..
"The driver seems to limit my hardwares performance to only 60% of what it was..."
-- I'm the root of all that's evil, but you can call me cookie..
Not only that, but I'm sure portions of nVidia's drivers also contain 3rd party IP as well. For them to open source that would be a breach of contract and would land them into legal trouble.
Life is not for the lazy.
4 years after they stop making the hardware, they finally mature enough to be relevant.
-- I'm the root of all that's evil, but you can call me cookie..
Being open means that these drivers won't simply go away once the product line is deprecated in favour of the newest and coolest graphics card, and that it will be able to receive improvements and bug fixes essentially until the last working piece of hardware dies off.
Wewt! I can get speed improvements! Now, at their current rate or increase, it will only take 5 years for the driver to be able to perform at the same level as the proprietary driver.
Being open also means that it will be able to provide support for this Radeon graphics cards in other platforms besides the officially sanctioned ones, such as Windows and Linux. Being open also provides a way to provide competition for the people AMD employs to develop their official graphics card drivers, because if an open driver developed by amateurs on their spare time happens to be nearly as good or even better then they may as well be out of a job, and they can't have that.
If, by 'competition' you mean something which performs sub par to your own drivers, I guess there's a point there somewhere.
Being open also means that, if the open drivers mature enough so that they are comparable to AMD's official offering, then it will be in AMD's best interests to get directly involved in the development of these open drivers and even abandon their proprietary offering in favour of this project. And, obviously, if these open drivers represent a business success story to AMD then you can bet that this will spread out to other companies, and everyone who used windows and had to deal with hardware with support problems certainly knows what a PitA it is to be tied to proprietary drivers which are crap.
So, a product which is developed for years and has only recently achieved 60% of what the commercial driver can do *isnt* crap.
-- I'm the root of all that's evil, but you can call me cookie..
The NVIDIA closed source drivers are miles ahead of ATI's closed source drivers as well as the open source drivers. NVIDIA cards are capable of decoding HD content in Linux while my four year old ATI card still only gets about 3 seconds per frame for 720p files with either driver. Also, the open and closed ATI drivers are buggy as hell, especially when running dual monitors. I bought an ATI card over NVIDIA because ATI released the documentation for their hardware, but unless there is significant progress on the open source drivers in the near future (especially in video decoding), then my next card will be NVIDIA.
You *really* didnt just make a comparison between CAD software..
And Mindcraft.......
-- I'm the root of all that's evil, but you can call me cookie..
While you are correct, you are not informative... Chill dude, stop being elitist and give the guy a break, he doesn't know.
For the GP: Basically CAD is used for design that will generally end up in a real world object. You can design for example a fancy box or even a desk/chair/house/skyscraper and then from your printouts (including the precise measurements) make the real thing. While you could do this in Blender, it is the wrong tool for the job. They're designed for modeling the real world not making something from scratch. Basically in Blender, you want something that looks like the real thing. In CAD, you want something that will help you build the real thing. They don't allow (easily anyway) the sort of precise pedantic precision you need for CAD. I have used both and I do know the difference, even if it is a little subtle. Never used maya, but I doubt it counts as CAD.
I have determined that my sig is indeterminate.
Not fair. RS690M's graphics may be garbage, but a laptop is a small sealed box full of discrete components, any of which could be causing the problem. In fact, the sleep/resume issue and graphics corruption, combined, look like a bad BIOS, which would agree with Gateway being a cheap/lousy OEM.
The preceding comment is my own, and in no way construes an opinon of the Emperor of Mankind.
Careful, you may invoke APK
Personally, I'd love to see some kind of APK/Dr. Bob crossover. Maybe about how maintaining a good HOSTS file prevents subluxations.
They say a little knowledge is a dangerous thing, but it's not one half so bad as a lot of ignorance. - Terry Pratchett
Nvidia claims that a significant amount of their graphics technology IS the driver and that opening it up would expose too much of their IP
For many, many years now NVIDIA has claimed they don't own the IP to large chunks of the code in their drivers, therefore it can not legally be exposed.
The truth of the matter is, NVIDIA's drivers, more often than not, have proved to be very reliable and fast even while ATI was openly bragging non-Windows platforms is for suckers. Beyond that, 99.999999999% of the world doesn't care if the driver is proprietary or open source so long as it performs well and is bug free. NVIDIA easily qualifies. And simply put, ATI's drivers still tend to have far more bugs and compatibilities issues than does NVIDIA.
Furthermore, I've been personally bit by ATI obsoleting GPUs twice. My brother once. I've never one had that bite with NVIDIA. When I dropped ATI I've never once seen a reason to regret it. Inversely, I've constantly seen reasons over that span where I'm extremely glad I gave up on ATI's "sucker bet" a long time ago.
The headline is very misleading. The big news here is that there is a Gallium3D back-end that gives tolerable performance for modern AMD cards. Gallium3D is a still-immature hardware abstraction architecture which could do amazing things for the industry if it manages to get over its chicken-and-egg problem.
Being open also provides a way to provide competition for the people AMD employs to develop their official graphics card drivers, because if an open driver developed by amateurs on their spare time happens to be nearly as good or even better then they may as well be out of a job, and they can't have that. Being open also means that, if the open drivers mature enough so that they are comparable to AMD's official offering, then it will be in AMD's best interests to get directly involved in the development of these open drivers and even abandon their proprietary offering in favour of this project.
TFA clearly states that the majority of the code was released by AMD devs and that they have recently expanded their open source driver team by adding two new members. Are you suggesting that this was all work done by the community? This is definitely not the case. Also, I would imagine that the Gallium3D driver is Linux-only and wouldn't be portable to other platforms, so they would still need the proprietary drivers for Windows.
When you think about it, if support gets better, then Linux could actually have even better AMD graphics support than Windows if Linux has the open option and Windows is stuck with Catalyst.
I have left slashdot and am now on Soylent News. FUCK YOU DICE.
This is probably a lot more to do with the fact that Maya performs faster on Linux, than it being a noticeable performance improvement related to the graphics driver. Its not the only app where this is the case, Houdini is also a tad faster on Linux than it is on windows. Linux is.. well... faster on the whole isn't it ? This may also probably the case for minecraft too. Also, I don't think Maya officially supports ATi cards, just saying that cos YMMV depending on the generation of ATi card too. Basically, there are too many vagaries to make anything approaching a meaningful comparison here. The only thing to do would be to run graphics based benchmarks on both platforms, assuming that the benchmarks are coded in the same way too of course. But they've already done that, hence the % figures in this articles title. As for the argument that CAD applications give the graphics cards more work, I really don't see how thats the case, a 3D viewport is just a 3D viewport after all. However, yeah, the processing in the backend to maintain the surfaces accuracy (is the term Type A surfaces ?), yep, thats a lot more hardcore than it is for 3D animation/modelling/rendering applications. But, and I'm interested to know if this if its wrong and someonecan explain it to me, drawing the 3D representation on screen isn't any much more of an effort for a CAD app than it is for something like Maya.
As long as they PROVIDE the driver, I don't care either. The whole free driver thing started because they didn't provide, and we didn't have a choice. Either you hacked a driver together, or you were SOL.
Today, companies do much better.
I8-D
The binary driver cannot be redistributed with the linux distros..
I've never had a problem with this. I always go to nVidia's or AMD's web site for my video drivers (on Windows). And some Linux distros DO provide binary drivers. I'll take a high-performance proprietary driver any day over a "free" but ineffective alternative. That said, I did pick one of my recent machines because it had a driver supported intrinsically by XOrg. Choice is good.
The binary driver may drop support for older hardware at any point, and the older versions which still support your hardware are unlikely support current kernels or X11 versions.
The only real argument you have, and a very good one at that.
You cannot fix a binary blob driver yourself, you are beholden to the vendor to do so.
Oh please. I doubt even 99% of Slashdot readers could fix a bug in a video driver -- particularly one for complex and high-performance hardware like nVidia and AMD video cards. Let's just put this old lie to rest, shall we? I have real work to do; I don't have time to be futzing about in some hardware driver for free, just for "free" software people can feel sanctimonious.
Also that "100%" is relative to the binary driver itself, its possible that given time the open driver will surpass it.
Possible, yes. Likely, no.
All about me
...it's only advantage is being Open?
I can see how many people may not see a great cost/benefits ratio there...
Nice troll. You could also say "the only advantage of living in a democracy is being free".
Have you got your LWN subscription yet?
Many reasons..
The binary driver cannot be redistributed with the linux distros..
Gimp doesn't distribute with the most popular Linux distro. Not all of it is license. Some of it is choice, and the fact that CDs are only so big. And this fix is quite simple if it can boot to VGA, which nvidia can. ATI could not for a long time...
The binary driver may drop support for older hardware at any point, and the older versions which still support your hardware are unlikely support current kernels or X11 versions.
You can still run a GForce2 on 11.04, so I do not see this as a problem, but a "potential" problem. Some people call that FUD.
You cannot fix a binary blob driver yourself, you are beholden to the vendor to do so.
Most users can not fix ANY driver themselves. And open source projects have lost interest and dropped support too... Admittedly, it is a strike against... Even a big strike, since no one can peer review the code. But it is far from a deal breaker.
Also that "100%" is relative to the binary driver itself, its possible that given time the open driver will surpass it.
Out of interest, does the open driver support OpenCL yet?
True, competition is good.
But what ticks me off is how fickle people are. Don't get me wrong; I am a big FOSS supporter, and involved in several FOSS projects. But I am not a purist... Nvidia was first to the party. When NO ONE was supporting Linux, they had a solid driver, with real support. It was even current! Now we have this new player at the party who ignored FOSS for almost all of it's history. Yes, they have a slightly more open license. They also have less people working on making a solid driver... But the fickle fanbois are ready to dump the one that has stood by Linux longer than almost anyone... Not me. I remember the heroes and villains longer than a year.
Oh, and Nvidia works better...
Yes, you really are I'm afraid.
You just said the equivalent of "Powerpoint is good enough for DTP right, so why worry about slowness in InDesign?"
Although that's not fair, actually, since Blender and Maya are actually pretty powerful in their own forte, but they're simply not CAD.
Because 3D horsepower is only used for games, right?
Don't you have homework to do before you're allowed computer time?
My experience with closed source linux drivers is that they're usually very poorly integrated with the rest of the system. The companies usually like to solve everything their own way (tm), rather than using the frameworks all the open drivers use.
When AMD dropped their support in fglrx for my radeon x1300-based GPU in my laptop (yes, they drop support for hardware whenever they feel like it) I had to start using the radeon driver on my ubuntu-machine. Everything has worked much better in the system since the switch. Suddenly I don't get some special AMD catalyst control center-thingie to change resolutions and set up external monitors etc. Instead the normal standardized gnome settings work like a charm. Also the system sets the correct resolution for my screen once, right after the kernel has been loaded (ie. before gdm/X).
If the radeon driver from TFA gets included in ubuntu 11.10 I would definitely give it a shot for my desktop machine, which has a Radeon HD 6870 card. The fglrx support for this card is just terrible. Sure, performance wise the OpenGL works well in games, but the normal X11 2D acceleration is terrible. Here are some annoyances with it:
* Whenever gksudo is activated it throws random garbage on all my monitors for about a second before displaying the password dialog
* Random "holes" in windows at random times, ie. squares where the desktop background suddenly becomes visible instead of the window contents. This won't go away until the window is redrawn.
* OpenGL and XV surfaces are always on top. So if I watch a video and put some window on top of the video surface, the video will be in front of the window regardless of Z-order.
If I could get a driver that plays nice with the rest of the OS, gets regular updates with the rest of the system, and doesn't have weird bugs in its 2D rendering I would gladly sacrifice 50% OpenGL performance. It's not like I utilize the GPU that much anyway.
When will NVIDIA change ways?
When the open AMD driver gets to at least 100% of the proprietary driver's speed in ALL cases, AND when AMD's sales start jumping up because of it, AND when Ubuntu decides to stop making it easy to install the Nvidia proprietary driver*. In other words, when they have a market reason to change ways and when competition is threatening to creep up on them if they don't.
Wow, that was easy enough. Next question?
*: Face facts, among Linux users who are going to want to use high-end video cards, the vast majority will use Ubuntu.
Demanding constant attention will only lead to attention.
Ok, I get it now.
Thanks for the explanation.
My book: Friendly F#, fun with game development and XNA; my game: Galaxy Wars by VSTeam; my gamedev language: Casanova.
Yes, but with the willful disregard for battery life the current trend in cell phone design, you'll be just as tethered to an outlet for your cell phone as you are to the 360 when gaming. I mean, yeah, great, you can run Doom 3 on a cell phone and connect it to an HDTV with HDMI, but can I just have the inexpensive, quick-enough option that lets me perform basic smartphone duties and has a battery life of two days with light to moderate use?
sleep/resume, thanks to ACPI, is a mess. A recent kernel release changed the behavior regarding trusting ACPI about being able to power down a device or not, and ended up draining batteries faster then earlier versions that insisted on powering down everything.
Sadly neither way is optimal as a device may have a legit reason for not powering down.
This however is less of a issue in Windows as the OEMs can work around flaky ACPI data in their drivers.
comment first, facts later. http://chem.tufts.edu/AnswersInScience/RelativityofWrong.htm
Some people can notice the flicker at 60 hz even. Going up further than that, a simple comparison test may yield no discernible difference. But unconsciously it can definitely make a difference. Remember, your brain doesn't see things, it interprets things. So while consciously, you may not be able to pick out a difference, higher quality video and audio still make an experience more natural and let you feel more immersed.
The highest quality displays can look 3d without any special effects simply by displaying an image so perfect that your brain literally thinks its the real deal. For decades, videophiles have been grading calibrated crt televisions based on the amount of depth they feel that they can see into the picture.
Not to mention high frame rates me a video card is ready to tackle bigger projects in the future, representing an investment.
---------- Open Source is capitalism applied to IP.
While I was dickish, I have neither the time nor the ability to even begin pointing out the differences between working in CAD to design something and "using a program that looks like CAD to design something."
The reference by the other person in this thread to Minecraft was insulting, frankly.
Just because you're working in 3D doesn't mean what you're doing is CAD.
I'm also going to go out on a limb here and say that Maya and Blender, while resembling parametric solid modeling to the casual user, aren't.
If this sounds elitist, then so be it. CAD is for stuff that eventually becomes physical. Accurate representation is /everything/.
--
BMO
Assuming a 7000.000.000 world population, your 99.999999999% would be 7 individuals. You really think only 7 people care about open drivers?
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
ATI cards have traditionally been better at video and dual head than 3d. Nvidia, traditionally, has been the exact opposite. Open source also drivers get a lot more attention to 2d graphics than proprietary drivers though, as everyone benchmarks and buys based on 3d performance.
---------- Open Source is capitalism applied to IP.
And yet I don't have these kinds of problems with laptops with nVidia, and I've had laptops with basically everyone's chips in 'em. Now, I DID have a different kind of problem with a poorly bonded die, which was a known problem. nVidia admitted it, and HP didn't. I hate HP more than ATI... But still resolve to avoid ATI.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
I said
>Accurate representation is /everything/.
I can't leave it at that.
CAD is also used to generate code for machining in post-processing. If i do not have a really accurate model, I'm not going to have parts that come out to the tolerances I want.
How accurate are the surfaces generated by Blender or Maya? Do they even close? For film and graphics and such, this doesn't matter. Nobody's going to care. A milling machine or lathe with garbage code from garbage surfaces, is going to give you garbage parts.
--
BMO
I understand the benefit of having a FOSS driver, but why would anyone in their right mind buy a $300-500 video card, then cripple the shit out of it by using the FOSS driver when their exists a proprietary driver that will give you 100% of the card's potential performance? It seems to me that instead of targeting the top-of-the-line GPUs, the Gallium3D project should be targeting the older GPU's that are more at risk of fading into obscurity. Yeah it's great they they can get the cutting-edge GPU's working, but it seems to be it would be better time spent working on better GPU support from the older cards that either are, or are soon-to-be unsupported by AMD.
Mommy told you to play nice.
And who said I ran 3D graphics hardware on Linux? My only Linux machine is an old G4 Powerbook.
Oh right, you thought I was "personally defending" myself or something, rather than just participating in a discussion. Ah, to be that narrow minded again! Simpler times!
The most important part in my view is that it allows for innovation in the open source graphics world. For instance, NVidia and ATI both make drivers for X.org however neither will produce drivers for Wayland until it gains enough traction. However Wayland will never get widely used until people can actually use it, which requires drivers. Open drivers breaks this stalemate.
What ever that can be done in hardware, can be done in software. The reverse is also true.
That totally depends on who buys NVIDIA after Intel and AMD squeeze them to death. At that point, the answer could be anywhen between "immediately after that" and "never."
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Until 51% of the people who voted decide to take away your freedom.
I'd rather have a constitutional form of government that guarantees my freedom regardless of what the majority decides. That of course implies democratic processes, but these aren't the ones making me free.
Mind the frickin' laser...
So you're saying I should use the proprietary drivers for the next X years (because they're still better), and then switch to the open source drivers when I don't have a choice anymore?
It's not like using the open source drivers contributes to their development either (unless you're actually finding and filing bugs... but if you keep running into bugs, why are you using the drivers again?).
you can get the same by enabling tear-free in catalyst or turning on vsync.
The nVidia vs AMD/ATI open source driver issue is all because hardware manufacturers pay for the patents in OpenGL. If you consider the driver part of the patent licensing, as nVidia does, there is no way they can open source the driver. If you consider the driver and hardware separate, as ATI does, you can open source the driver, with some caveats - namely it can't be called an OpenGL driver (it is OpenGL compatible). WINE operates similarly as a Windows compatible API.
Personally I've found nVidia's extension support has traditionally been much better than ATIs on consumer hardware, though ATI promised changes to that a couple of years ago (my newest graphics card is 2 years old and nVidia, so I haven't checked it).
Possibly, but not with Compiz or KWin. Believe me, I've tried every combination of sync and rate in both the compositor and driver and, until GNOME3, none of it worked unless compositing was turned off.
Watch the "elephant charge" or "crack of doom" scenes in Return of the King. It's gotten better over the years, but this is the first time I've seen video playback on Linux on par, well, with what Mac OS has been doing for a decade. I don't know what the Mutter devs did, but they did it right.
--srj/mmv
I'd rather have a constitutional form of government that guarantees my freedom regardless of what the majority decides. That of course implies democratic processes, but these aren't the ones making me free.
And neither does free/libre software directly make you free, but it is an important link in the chain. See the poster in this thread who runs ATI hardware on hardware not supported by AMD. Though it was the effort of programmers not employed by AMD that directly made that come about, however this most probably would never have happened without software freedom.
Have you got your LWN subscription yet?
Point taken...
Cad programs typically use double precision coordinates. Maya/Blender use single. Most video cards doubles run at 1/4 speed of single precision numbers even with the proprietary drivers. Fermi based Quatros being the only exception I am aware of. I wouldn't be surprised if the free drivers were worse than that.
Cad is the only area where this particularly is a big issue (unless you count games with really big world sizes).
Maya & Blender are considerably more than generic desktop stuff however.
Wewt! I can get speed improvements! Now, at their current rate or increase, it will only take 5 years for the driver to be able to perform at the same level as the proprietary driver....So, a product which is developed for years and has only recently achieved 60% of what the commercial driver can do *isnt* crap.
Dear troll, no it is not crap, it is running on my 4 way Phenom box right now, very nicely. I had problems with the Catalyst driver including odd behavior on reboot and horrible breakage on nearly every kernel upgrade. When I have time on my hands I will fiddle with the Catalyst driver some more and get it working again, it does have advantages. But using it does entail a certain degree of pain I am not willing to suffer at the moment. I regard the Calalyst driver as a hobby project, the open driver as my reliable work horse.
Have you got your LWN subscription yet?
Actually, these benchmarks are for the 6000 series Radeons. So, 6 months after they started making them.
Why aren't you going to be doing any gaming? Because there aren't many games that are made to run well on Linux. And why is that? Because the graphics stack sucks for 3D. It's a circular argument that has allowed the state of Linux graphics to continue to lag behind embarrassingly. I can't see how ATI being committed to improving their Linux drivers is anything other than a good thing?
99.999999999% of the US didn't care that houses were being sold for far more than they a healthy housing market could bear. They didn't care that selling their house at rediculous prices and taking part in a totally unsustainable market bubble. People are short sighted and greedy. They are happily take short term gains that screw themselves in the long run.
Claiming that proprietary drivers are better because they work better right now is exactly that kind of short sightedness. AMD's move to help get open source drivers working on their hardware is specifically to prevent situations like yours where you you got bit by obsoleting GPUs. Those drivers that run at 60-70 percent of speed means that while you might take a hit, you won't be left out in the cold again. I have been bit by obsolete drivers from both ATI and nVidia so I am cautious either way.
AMD is taking the sustainable long term approach. nVidia is taking the short term more profitable approach. AMD has acknowledged their problems and are working at fixing them. nVidia is still at the party and has not yet had to deal with the falloutl
Fair enough. I have an HP Elitebook 8530p with Radeon Mobility 3650, and it works perfectly, 3D, 2D, sleeps and wakes, under Gentoo and Win7 both, so we'll just have to be opposing sides of the anecdotal coin.
The preceding comment is my own, and in no way construes an opinon of the Emperor of Mankind.
It still fails on the high precision side of things. Can you create an object and produce highly accurate detailed labeled standard drawings in blender easily? If the answer is know, then blender qualifies as Computer Inhibited Design, not CAD... The differences I highlighted were fundamental, but by no means a complete list... The idea was to give the general idea not the 3 year university degree... :)
It is one thing to make your model "real world" and another to publish precise measurements so that other objects can interlock with it... If you really pushed it, I suppose you could use blender for industrial CAD, but it would be long, slow and generally problematic. Wrong tool for the job really. But it does make nice 3D logos and animations and stuff, which are much harder to do with proper CAD software.
I have determined that my sig is indeterminate.
Fair enough. I wasn't suggesting you give him a full course(and my explanation wasn't intended to be complete either), but maybe rather than a simple "yes" a bit of "well here is something to get you started on understanding the difference..." might have been helpful to the poster...
I have determined that my sig is indeterminate.
... If the answer is know...
* no.
I cringe seeing that!
I have determined that my sig is indeterminate.
Ouch, indeed. I blame copy/paste.
Ahhh...the great dumpster continuum. Many a free computer will be found there. -- sowth (748135)
There's also the worthwhile argument of power management - the open drivers don't have very good power management yet. Especially for Nouveau, where power management on almost every card is still a work in (rapid) progress. And it's complicated too, because you have to read information from the VBIOS, and half the time the VBIOS is broken in some way or another. And there are too many for the developers to test all of them.
The proprietary drivers, on the other hand, have generally working power management for most all cards that they support.
So what did you use a few years ago, when your 2D and 3D acceleration were slower than the open drivers are today?
will never open up fully. And Microsoft will by up their IP at the fire sale to fuck over Linux just a bit more.
By the way, if you want free 2D Cad for Linux, get your ass over to Dassault Systems and download Draftsight.
I'm logging in and commenting for the first time in years to say thank you for sharing this information. I've been dying to find a decent CAD package for Linux.
Being open means that these drivers won't simply go away once the product line is deprecated in favour of the newest and coolest graphics card, and that it will be able to receive improvements and bug fixes essentially until the last working piece of hardware dies off.
Wewt! I can get speed improvements! Now, at their current rate or increase, it will only take 5 years for the driver to be able to perform at the same level as the proprietary driver.
Um, no. The 3D driver in question is the R600 Gallium driver, which only started development a year ago. So at the "current rate of increase", it would take a few more months to reach the same level as the proprietary driver.
Being open also means that, if the open drivers mature enough so that they are comparable to AMD's official offering, then it will be in AMD's best interests to get directly involved in the development of these open drivers and even abandon their proprietary offering in favour of this project. And, obviously, if these open drivers represent a business success story to AMD then you can bet that this will spread out to other companies, and everyone who used windows and had to deal with hardware with support problems certainly knows what a PitA it is to be tied to proprietary drivers which are crap.
So, a product which is developed for years and has only recently achieved 60% of what the commercial driver can do *isnt* crap.
Correct. A product which has been developed with maybe 5% of the manpower of the commercial driver, has better 2D performance, is more stable, and has 60% of the 3D performance is not crap,.
Being open means that these drivers won't simply go away once the product line is deprecated in favour of the newest and coolest graphics card, and that it will be able to receive improvements and bug fixes essentially until the last working piece of hardware dies off.
I wish that was true - but unless I'm being misled, these drivers already don't support my 2-year-old card or the generation after it. Is there anything concrete to give me hope this will change, and the 6000 series they're now making great steps forward with will be supported for more than 10 minutes after the 7000 series is released?
Wow. You think the proprietary AMD drivers are perfectly optimized to run at 100% of the hardware's limit? That is a lot of faith to put in AMD.
The problem with your statement bubba is the fact that I just bought a replacement system in March that includes an HD4220 Radeon IGP with no PCIe Video Card Slot. This is a business system so upgradeability is limited. As a small business owner, I'm not willing to replace a working system for at least 5 years as the IGP is sufficient to handle the Win7 Aero Needs and such with reasonable performance. Keep in mind this system was not bought for Gaming/Fun. It's a Business machine, thus makes money.
Mod me up/Mod me down: I wont frown as I've no crown
Blaming the GPU for something the BIOS has the most to do with. #rootcausefailure
The Gallium3D driver has been under development for nearly 4 years. Out of the box they where able to get roughly 30-40%, so, after 3 years, it increased roughly 10% per year. So yea, for 100%, itd be about 5 years. It should be noted that it performs in selected tests at that speed, NOT across the board. The more complicated and graphics rich the test, the worse the open source driver does. For example, the Lightsmart benchmarks come in at around 25%.
-- I'm the root of all that's evil, but you can call me cookie..
....
'60% of what it was using the proprietary driver'
way to deflect the point..
-- I'm the root of all that's evil, but you can call me cookie..
They have NOT achieved what people are saying they have. the 60% number is a best case, with one or two specific benchmarks. In others, they are down by 25%. And they had achieved most of the initial speeds simply porting the old driver over, due to the similarities in the architecture.
-- I'm the root of all that's evil, but you can call me cookie..
But somehow, the catalyst driver *IS* crap, which is what I was responding to. Only in the last month did they even consider the open source driver ready to be USED without crashing and acting strange.
For graphically intensive environments, many people have had the exact opposite experience, that the open source drivers require more time and effort then the proprietary ones.
-- I'm the root of all that's evil, but you can call me cookie..
You can thank not having a stable kernel ABI for that - so it's more like "the kernel not playing well with proprietary drivers" than it's "the driver not playing well with the kernel".
Sometimes I wonder if one of the main reasons for not providing a stable kernel ABI is to make life hard for proprietary drivers, rather than "being able to optimize stuff".
Then again, ATi is notorious for making bad drivers. One of the only times I've had really bad crashes on Windows (after moving to NT) was because of ATi... and that was spectacular enough to result in two complete losses of data. Hurray backups!
Coffee-driven development.
Other, only one. We see that 50-60% was pretty normal for the gaming benchmarks, it was just the workstation benchmark that suffered, probably due to low interest (high performance workstation activities lie outside the scope of "good enough for most"). And there is a relative % improvement over the performance of the 5000 series (using that "old driver" you mentioned), which I think is very notable.
Only one when your just looking at that one singular article. Across the board, the Gallium3d drivers perform at around those levels across most of the spectrum.
I'm not saying it isn't notable for what they've accomplished. I'm just saying using this as an example of how proprietary drivers 'suck' is just silly. I, personally, would certainly use the open source driver, if the proprietary driver was failing for me. But jumping to the open source driver 'just cause' doesn't make much sense. Unless, of course, Linux is going to suddenly suck because the license isn't as liberal as hurd.
-- I'm the root of all that's evil, but you can call me cookie..
I understand that most users lack the skill set to fix a driver. Some people do possess this skill however and if the driver is open there is a chance someone will take up the challenge to fix it. If the driver is proprietary there is no chance whatsoever.
Yeah... I guess licenses like the GPL, Apache and what not are the 'constitution' in this analogy.
Mind the frickin' laser...
I'm with you on this one. NVidia did produce a linux driver when ATI spit on linux users. I haven't bought a single computer with ATI since NVidia started producing a linux driver. I remember how grateful I was to actually have the ability to use my graphics cards 3d capability. I tend to hold grudges a long time too.
There is no, and most likely never will be any "ABI" inside the kernel. This is a design decision that has nothing to do with proprietary drivers and everything with improving interfaces between components -- source compatibility stays the same, and proprietary drivers' vendors can accommodate their blobs with wrappers. It is also unrelated to the fact that all proprietary drivers except Nvidia, are steaming piles of shit that no one wants to use.
Linux ABI is its syscall and library interface -- it's very much backward compatible, and proprietary applications use it just fine. If even Adobe developers figured out how to release Linux binaries, I can't see how anyone can complain about it.
Contrary to the popular belief, there indeed is no God.
when you write software/drivers you cant expect the first, second or even third version to be perfect. you make it work and then you optimize it. this is the same way the proprietary drivers started out.
Anons need not reply. Questions end with a question mark.
The binary driver may drop support for older hardware at any point, and the older versions which still support your hardware are unlikely support current kernels or X11 versions.
You can still run a GForce2 on 11.04, so I do not see this as a problem, but a "potential" problem. Some people call that FUD.
Actually, I have a Geforce2 Go in a very old laptop (from 2001, Pentium III based), and suspend or hibernate never worked with the blob, and it never will, because the blob is only updated to support new interfaces. In its time, when I went to travel, I usually used the nv driver to be able to suspend and hibernate properly. Nowadays I can use nouveau, but that driver doesn't support 3D (at least not yet).
Nvidia was first to the party.
Not really, the early Matrox cards (like the Mystique, G200/G400) had nearly complete FOSS support, only dual head with video out needed a proprietary library in the beginning, I think it was because of Macrovision. Later, with the framebuffer device one could actually use the second screen without this blob. That was arond 1998 when NVIDIA promised FOSS drivers and dumped some obfuscated code on the utha-glx project. Then they started to give out blobs, and they where buggy - at least when you wanted SMP support. 2002 someone sponsored the development of FOSS drivers for the ATI R200 series cards, I think the base for the good R300 support also stems from that time. It was only around that time, when the NVIDIA drivers finally became stable on SMP.
Of course it is also true that there are no FOSS drivers for newer Matrox cards and the closed ATI drivers sucked big time until AMD stepped in. When they announced their FOSS strategy in 2008, I was just putting a new computer together and so I decided to get an ATI-AMD card to support the cause. I use the blob, and one could really see it improving significantly since.
When NO ONE was supporting Linux, they had a solid driver, with real support. It was even current! Now we have this new player at the party who ignored FOSS for almost all of it's history.
It is true that there was a time, when NVIDIA was the only reliable option when you wanted decent 3D on Linux, but AMD was always supportive of FOSS and Linux, and when they bought ATI, they proved it by starting to make their blob usable, and also by supporting the development of free drivers.
The Gallium3D driver has been under development for nearly 4 years. Out of the box they where able to get roughly 30-40%, so, after 3 years, it increased roughly 10% per year. So yea, for 100%, itd be about 5 years. It should be noted that it performs in selected tests at that speed, NOT across the board. The more complicated and graphics rich the test, the worse the open source driver does.
4 years? Where are you getting that number? The initial commit in Git of the r600g driver was in May 2010. May 2010 to July 2011 is not even close to 5 years.
For example, the Lightsmart benchmarks come in at around 25%.
That's not the Radeon driver's problem. All Mesa-based drivers have had similarly poor performance in Lightsmark for a long time, which suggests a shared CPU bottleneck.
I think you're being misled. Even R300 cards much older than 2 years are still supported.
Even though the FOSS driver stack doesn't use either of those licenses. The entire stack was MIT/X11-licensed last I checked, including the kernel parts.
4 years? Where are you getting that number? The initial commit in Git of the r600g driver was in May 2010. May 2010 to July 2011 is not even close to 5 years.
The r600g driver was reworked based on capabilities present in the r300 driver.
For example, the Lightsmart benchmarks come in at around 25%.
That's not the Radeon driver's problem. All Mesa-based drivers have had similarly poor performance in Lightsmark for a long time, which suggests a shared CPU bottleneck.
Actually, another way to look at it is, all Gallium drivers exhibit similarly poor performance in intensive tests/use. But the point is, it's an alternative driver. Great! But it doesn't mean it's better. People buy hardware for performance, specifically, graphics cards. Perhaps the performance will someday surpass that of the proprietary driver. But it's not exactly newsworthy until it is at least equal or surpasses.
-- I'm the root of all that's evil, but you can call me cookie..
But somehow, the catalyst driver *IS* crap, which is what I was responding to
Overreacting by attacking I would say. The Catalyst driver is not crap either. I have used it a lot. Its binary installer on the other hand is poor. It will neither install the Catalyst driver completely nor remove it completely, leaving you in for considerable pain especially if you do not know your way around xorg.conf and especially is you do not have a second, web-connected machine at hand to track down problems while your GUI is broken by your attempt to install catalyst.Stupid, preventable issues like mysterious driver segfault if you have run the binary installer but not yet run the control center utility. Also, segfault if you have attempted to remove catalyst and not hand edited your xorg.conf to go back to the xorg driver. You may also find DRI failing to initialize and glxinfo just segfaults uninformatively.
Sticking with the xorg driver(s) is the best way to avoid this and other pain. The vast majority of users do not need more than a fraction of the horsepower even the cheapest Radeon delivers these days. By that measure, 60-70% of the proprietary driver performance is overkill. Even the factor of three or four difference we saw a few months ago is unlikely to be noticeable except in things like high end shooters at high resolution.
It's nice to have options, don't you think? To be sure, the driver differences are not just speed, certain capabilities of the card are not implemented as well in the xorg driver as in catalyst. As of xorg 7.6, lines are antialiased by catalyst but not by xorg (3D games typically draw thin polys instead of lines so this only affects things like engineering and game development applications). Mipmap filtering is visibly better in catalyst. I am sure this is just the beginning of the list. Nonetheless, the typical user just isn't going to notice, the xorg driver is already good enough.
As for me, I am happy that both open and closed drivers are actively developed, the latter as a cooperative effort between AMD and many outside developers, both professional and volunteer. The open driver will eventually surpass the closed one if history is anything to go by, and not just for products that have been on the market for years. Kudos all round I say, two code bases would seem to be better than one.
Have you got your LWN subscription yet?
The current status is here: http://cgit.freedesktop.org/mesa/mesa/tree/docs/GL3.txt When all is supported, the version of mesa will jump to 8.0. this month 7.11 will be released, and the next release is to be expected in January/2012.
But somehow, the catalyst driver *IS* crap, which is what I was responding to
Overreacting by attacking I would say. The Catalyst driver is not crap either. I have used it a lot. Its binary installer on the other hand is poor. It will neither install the Catalyst driver completely nor remove it completely, leaving you in for considerable pain especially if you do not know your way around xorg.conf and especially is you do not have a second, web-connected machine at hand to track down problems while your GUI is broken by your attempt to install catalyst.Stupid, preventable issues like mysterious driver segfault if you have run the binary installer but not yet run the control center utility. Also, segfault if you have attempted to remove catalyst and not hand edited your xorg.conf to go back to the xorg driver. You may also find DRI failing to initialize and glxinfo just segfaults uninformatively.
Sticking with the xorg driver(s) is the best way to avoid this and other pain. The vast majority of users do not need more than a fraction of the horsepower even the cheapest Radeon delivers these days. By that measure, 60-70% of the proprietary driver performance is overkill. Even the factor of three or four difference we saw a few months ago is unlikely to be noticeable except in things like high end shooters at high resolution.
It's nice to have options, don't you think? To be sure, the driver differences are not just speed, certain capabilities of the card are not implemented as well in the xorg driver as in catalyst. As of xorg 7.6, lines are antialiased by catalyst but not by xorg (3D games typically draw thin polys instead of lines so this only affects things like engineering and game development applications). Mipmap filtering is visibly better in catalyst. I am sure this is just the beginning of the list. Nonetheless, the typical user just isn't going to notice, the xorg driver is already good enough.
As for me, I am happy that both open and closed drivers are actively developed, the latter as a cooperative effort between AMD and many outside developers, both professional and volunteer. The open driver will eventually surpass the closed one if history is anything to go by, and not just for products that have been on the market for years. Kudos all round I say, two code bases would seem to be better than one.
Your points are valid. Here is one distinct advantage of the open source driver. It can be used to support environments outside of X eventually, hopefully assisting in bringing us out of X11 into a new environment which isn't as bogged down with legacy support.
-- I'm the root of all that's evil, but you can call me cookie..
Careful guys...
It might have rabies!!!
"City hall" in German is "Rathaus" Kinda explains a few things......
You seem to be using English words, but what your saying bears no resemblance to reality.
AHHAHAHAHAHAHAHA
Loser!!!
"City hall" in German is "Rathaus" Kinda explains a few things......
Um, no. The 3D driver in question is the R600 Gallium driver
The Gallium3D driver has been under development for nearly 4 years.
The Documentation for the R6XX has been released in December 2008 and the R6XX/R7XX programming guide in May 2009. It may be that the core Gallium 3D driver is in development for four years, but given the availability of the documentation, the R600 was developed for at most two and a half years. Actually, first support for the HD6000 series of graphics cards was added in January 2011.
AMD hasn't opened their drivers. What they did was release the specs and documentation for their GPUs and other people write the open source drivers. AMD still has it's own proprietary drivers which are faster and support more features than the open source ones.
Mada mada dane.
You've probably been mislead: This is the current feature matrix.
Oooooh! You can write a program to reverse a text string! I cower from the obvious supremacy of your 1337 programming skillz!!! HAHAHAHA I bet you pulled that off the Internet with Google, because you couldn't even figure it out yourself!
* Is THAT the "best you've got" troll? Because you surely didn't do to well in the next 2 links below... now did you? Nope... lol!
I love how the best you can do is link to your own posts, with my statements cut down and taken out of context, because that's the only way you can twist them to say what you need them to say.
Like when I supposedly "agreed" with you that your method of moving rootkits works. You conveniently left off the part of my statement where I tell the type of rootkit it doesn't work against, because that's the only way a loser like you can feel like you won an argument.
What? Do you just hope that people won't read back in the thread to see what I actually said? Everybody on this board knows your too much of a douche to take anything you say at face value.
HAHAHAHA
Loser!!
"City hall" in German is "Rathaus" Kinda explains a few things......
Ok...let's look at your twisted "summary" post:
Your CD you didn't write the tools for yourself..
Well, your first sentence is an outright lie, so what does that tell us about the rest of your post?
In your paranoid delusions, you made the assumption that I didn't write it myself, because that would have conflicted with your polka dotted sky worldview that you're the only one in the universe capable of turning a computer on. I just didn't correct you, because I was enjoying the complete ass you were making of yourself.
(Once I was pointed out your CD wasn't needed & obsolete against that rootkit, YOU got pissed off!
Pissed off? lolwut? I was laughing my ass off at the stupendously moronic things you were posting. I could just imagine every post, you pounding on your keyboard harder and harder, as your mouth is foaming in rage that I dare challenge your 1337 5k1llz.
Besides, I'm pretty sure it's you that's pissed. You're pissed that I can run the recovery console on a non-bootable computer remotely, without any special hardware or tech knowledge on the end users part, and you have no freaking clue how I do it. And that just burns you up, doesn't it?
"Will it get rid of an MBR rootkit? Yes. Will it get rid of a driver-based rootkit with a discrete .sys file for the driver? Yes." - by cbiltcliffe (186293)
There you go again, conveniently removing the part where I said it won't work against a rootkit that patches legitimate driver files. I wonder why you felt the need to do that? Maybe because you know I'm right, and leaving it in would cause a paradox in your head, making your brain (what's left of it, anyway) melt?
So your first three points have been blown out of the water. I'm not going to bother continuing, because it's just more of the same.
"City hall" in German is "Rathaus" Kinda explains a few things......
4 years? Where are you getting that number? The initial commit in Git of the r600g driver was in May 2010. May 2010 to July 2011 is not even close to 5 years.
The r600g driver was reworked based on capabilities present in the r300 driver.
Even if you count r300g (which you shouldn't, since their capabilities are different enough to warrant 2 different drivers), the first commit to r300g was in January 2009. That's 2.5 years, not 4.
For example, the Lightsmart benchmarks come in at around 25%.
That's not the Radeon driver's problem. All Mesa-based drivers have had similarly poor performance in Lightsmark for a long time, which suggests a shared CPU bottleneck.
Actually, another way to look at it is, all Gallium drivers exhibit similarly poor performance in intensive tests/use. But the point is, it's an alternative driver. Great! But it doesn't mean it's better. People buy hardware for performance, specifically, graphics cards. Perhaps the performance will someday surpass that of the proprietary driver. But it's not exactly newsworthy until it is at least equal or surpasses.
It is newsworthy because there are other aspects to graphics drivers besides 3D performance in benchmarks. Specifically, the open driver is more stable than Catalyst and has better 2D performance, and people care about those things too. Improved 3D performance, even if not quite up to the level of Catalyst, addresses the main remaining problem with the open driver. It doesn't necessarily have to have equal or greater performance to reach equal or greater quality for most users.
You mean like VDPAU or VA-API? I'll use the free drivers when they can use my cards support for in-card video decoding so I can play 1080P back with 1-2% CPU usage. Until then it's proprietary for me... especially on my NVidia cards.
To err is human; effective mayhem requires the root password!
If I were to design and 3d print a fancy paperweight in blender, is this not computer aided design?
Will I really care about tolerances on it when said physical objects only purpose is to look nice and be heavy?
Wholeheartedly agree that blender is ill suited for cad, and you aren't exactly going to design engine parts with it even if you do use it to design things, but to be pedantic it would still be computer aided design so long as something physical with the general properties wished for is made.
If Microsoft produced a driver that ran at less the the speed of a official proprietary 'Catalyst' driver, Slashdot posters would be rolling in the isles laughing and criticizing Microsoft for its failure to produce a quality driver. When the FOSS community does the same thing, it is heralded as the greatest thing since the invention of the wheel. However, as surely as day follows night, a poster(s) did involved Microsoft into the mix. The creed of the FOSS and *.nix/*BSD lovers everywhere, "When in doubt, blame Microsoft". It is easier to blame others than to get off your dead ass and do something constructive to correct a problem.
I will now yield the floor to Anonymous Coward who will thrill us with his vocabulary skills.
Pigskin-Referee
Linux: Yesterday's technology, tomorrow
I'd suffocate both of you if could. APK is nuts, he has an excuse, what's yours?
I know tobacco is bad for you, so I smoke weed with crack.
Less bugs? Though it gets attention from AMD, the proprietary is known to have a lot of random bugs, some of them even making Gnome 3 pretty much unusable. I've had more positive experiences with the open driver so far, aside from the fact that it won't run some 3D stuff.
The performance feels closer to 30% for me, though. Perhaps the Radeon HD 4330 uses a different driver or just has less implemented for it, still.
I am not devoid of humor.
Which means what he said doesn't make sense. If it did, then the free drivers would have been at 0% of the speed.
I am not devoid of humor.
I think it makes the point. How fully you use the hardware is a point on a scale of grays.
I happen to think APK baiting is fun?
You obviously don't see the humour in it. I apologize if you think it's a waste of /. database space.
Others have commented to finding it funny, so it's all a matter of perspective, I guess.
"City hall" in German is "Rathaus" Kinda explains a few things......
"out-of-town"
Interesting that you put it in quotes.
I think "out-of-town" == "from the mother planet"
"City hall" in German is "Rathaus" Kinda explains a few things......
Hey, apk. Nice to talk to you.
To answer your question, I'm only a jerk to people who are abusive themselves, amd have shown they have no chance of being useful for anything else...
Psychology 101 pop quiz time:
You think a bunch of people from around the world are actually 1 person attacking you with dozens of pseudonyms. Which of these is most likely the truth:
1. A bunch of different people around the world have conspired to attack you.
2. The different identities are in fact pseudonyms, and it is a single person who is attacking you using them.
3. You're consumed by paranoid delusions, and the people attacking you are actually a bunch of people simply making fun of you because of your delusional fantasies, which you continually foist on anyone who's not interested.
"City hall" in German is "Rathaus" Kinda explains a few things......
Talk about an offtopic post. I think you're in the wrong story here, apk.
You really need to think when you astroturf, apk. You're not fooling anyone. You're seriously trying to say that you're a random anonymous surfer who happened upon a thread in a 3 week old story, and then went to a completely different story to post a comment about that thread?
You only baited him with your immature taunts because he upset your sensibilities by pointing out that your CD was unnecessary for killing the unkillable rootkit using what most everyone has a copy of in a Windows installation CD or DVD
"Everyone" is a big word. It implies that all those millions (probably billions) of people who bought a name-brand system - with a recovery partition, rather than a CD - don't exist. It also implies that all those people that bought a system with Windows XP Pro Corporate VLK installed on it due to a "special deal with Microsoft" that the seller had, don't exist.
and that he had methods using process explorer that killed the botnet portion operating in usermode also. I read what was written and you tried to twist what he said into saying he was using process explorer and tcpview for detecting rookits. He never said that and you could not produce proof of he even implying it, let alone a direct explicit quote.
You included both botnets and rootkits in the same sentence, using Process Explorer and Recovery Console as the tools to eliminate these. There was no attempt to differentiate the two tools from each other as far as their use was concerned. Paraphrased, you said "When standard tools fail, for botnets and rootkits, I use Recovery Console and Process Explorer." I did quote this, and you chose to ignore it. I also quoted another statement which implied the same thing. You chose to ignore this, also.
Just because you choose to ignore evidence that's placed in front of you if it conflicts with your polka-dotted sky worldview, does not mean there is a problem with my argument.
You also later said he said to use process explorer to kill the rootkit. When he challenged you to produce a quote of him saying that explicitly you could not.
I never once accused you of saying process explorer would kill the rootkit. That's even too stupid for you to claim. I think I'd remember if you made that claim.
He said you have reading problems. I'd be inclined to agree, except for the fact your posting history shows you harping on hosts files and immediately afterwards sock puppets obviously of yours in alternate registered user accounts support you? Please. Give us a break. We're not that stupid.
Since you're reading my posting history, see my previous post on Psychology 101 pop quiz. I have two accounts on /. One of these is this one, and the other is...I can't even remember the name of it, as I haven't logged in to it for about a year. It's something to do with Jack Thompson and his crusade against video games, though. Again, the fact that everyone thinks you're a crackpot doesn't mean they're all sock puppet accounts. It usually means that you are, indeed, a crackpot.
And spare me the BS about "You need a psychology PhD and a proper exam done in controlled conditions, otherwise you're libeling me!!" "Crackpot" isn't a medical term. Neither is "nuts" which somebody previously used to describe you. They're quite obviously opinions, and opinion is not libel. It's protected speech. If somebody described you as criminally insane, you might have a case, but tossing around barbs like has been done so far is perfectly legal.
You point out that your cd allegedly can remotely do the same, but that's not proven, nor is it even known it can.
Seriously? You're scraping pretty low to come up with something against me if that's the best you can do. You haven't proven that you're apk, either. (I know, I know....
"City hall" in German is "Rathaus" Kinda explains a few things......
While I was dickish [..]
Apology accepted; just do not take the next one for granted.
I have neither the time ..
Try spending less time being a smartass- it works fine for me
.. nor the ability
It seems that you do have the ability, and a fine one I might add. You explained what CAD is pretty well, as did others, and now I know.
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
The first post you linked to, I've already refuted everything in it. Again, you chose to ignore it.
The second post is this little literary thing called "hyperbole". Maybe you've heard of it. Since you claim to be such a computer genius, maybe you know how to look it up using this little thing called Google. Maybe you've heard of it?
No, of course, to ALL of the above for you!
There you go again, making assumptions. My opinion of you as having delusions of grandeur is protected free speech. You claiming as fact that I don't have a PhD or experience in these things is libel if it's not true. Do you really want to go there?
The only person at risk of legal action here is you.
P.S.=> HOWEVER - The funniest one was your "1st attempt here" @ doing so, when YOU partially quoted me to TRY to save face, & only blew it once again too
Really? You're still trying to use that partial quote of mine to justify yourself? That's pathetic. It's already been completely refuted multiple times by me, so you haven't got a leg to stand on, and you know it.
Pot calling the kettle black again, Forrest? See here -> http://slashdot.org/comments.pl?sid=2326772&cid=36781858 [slashdot.org]
You're allowed to your opinion, and I'm allowed to have an opinion on your opinion. Just don't start trying to state things as fact, because it can get you into trouble. Especially when you seem to know less than you think, like you do, and have a hot temper to boot.
I also never said it was a "bunch of people" - it's just YOU, using sock-puppet alternate registered accounts
No, you're wrong. It's a bunch of people. Pretty much everybody on this board has a similar opinion of you. It's no surprise that the vast majority of them are agreeing with me. /. standards) story. Then hunting me down somewhere else to "tell me off" about what I said. I know it's you.
And as far as others having "told me off repeatedly". Really? They seem to know things that only you would realistically know. Like following a thread on an ancient (by
"City hall" in German is "Rathaus" Kinda explains a few things......
I'll switch when it is faster or more stable, but until then why limit my hardware?
Before nVidia supported Linux, there was good 2d support for earlier ati cards, s3 cards, cirrus logic and matrox... In terms of 3d, the 3dfx voodoo cards had pretty good support (they even provided glide drivers for linux), as did the matrox cards (they opened up their drivers, and the open ones surpassed the closed ones fairly quickly).
ATI i don't especially like, not only did they ignore linux users, they also have traditionally provided very buggy drivers for windows too... On the other hand, AMD have always been good to the community, and they now own ATI so I have no qualms about using their cards today.
A few years ago i would not have even considered buying an ATI card, however my 2 most recent videocard purchases have been from them, and it's good to know that the open drivers will continue to support these cards long after the closed ones have moved on.
http://spamdecoy.net - free throwaway anonymous email - avoid spam!
Unfortunately the /. comment system can identify who is behind the AC, and can't filter appropriately, thus cluttering up the thread visually - nothing personal, but -1, Flamebait exists for a reason.
I know tobacco is bad for you, so I smoke weed with crack.
You *really* didnt just make a comparison [..]
I obviously did not make myself understood; I mentioned Minecraft because it is popular (so many people would have something to say about it), it runs on many platforms (so comparisons between different OSs can be made), and its performance depends on your card and drivers- that last point, by the way, being what this thread is about.
Though I did not intend to mention Minecraft implying that it is CAD software (and apparently offend a bunch), in retrospect, I should point out that there are very many people who use it for art.
The three laws of thermodynamics:(1) You can't win. (2) You can't break even. (3) You can't even quit.
Why should I disprove what you wrote for a fourth time, when you haven't even addressed any of my points even once?
You've ignored all my repeated proofs so far, with your hands over your ears going "LALALALA I CAN'T HEAR YOU!!!", so why should I do the same thing again, just for you to ignore again?
"City hall" in German is "Rathaus" Kinda explains a few things......
Why don't they use something more modern like the Unigine benchmark, or a not so ancient game like Enemy Territory Quake Wars?
Because on the state tracker side of things, Mesa/Gallium3d doesn't support as much eye candy as Catalyst.
Gallium3d is slowly entering the world of OpenGL 3.x, whereas the proprietary drivers are at OpenGL 4.1
It will come with time (specially since Google made the opensource Intel drivers jump on the Gallium3d bandwagon), but for now you'll have to be patient.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
but why would anyone in their right mind buy a $300-500 video card, then cripple the shit out of it by using the FOSS driver when their exists a proprietary driver that will give you 100% of the card's potential performance?
When :
- you don't need 100% perf 100% of the time. Some people dual boot into Linux for work and into Windows for games.
- you want to vote with your wallet for a company which actively supports FOSS driver development. Even if you personally won't run the open-source drivers, and end-up for now running the blob, you're still encouraging a hardware maker which supports open-source.
- some time people cycle the hardware between several machine. Today, you'll be getting top of the line performance on your main machine with the catalyst driver. In a couple of years, this card might end-up populating a less important machine in your kitchen. By then, the performance and power-saving would have become better for this card and you'll be happy to have a driver solution which runs perfectly out of the box with Ubuntu Zanny Zebra when AMD have dropped official support from their BLOB.
it seems to be it would be better time spent working on better GPU support from the older cards that either are, or are soon-to-be unsupported by AMD.
- Today top of the line card are tomorrow's older cards. Because back then, when they were top of the line, people *did* work on supporting them in OSS, now the Radeon 9600 to Radeon X1950 generation of cards have a nice working r300g driver, which is also recommended by AMD as the official way to support these older cards.
- The same way, you might complain that it doesn't make absolutely sense right now. But, as written above, in a couple of years down the line, the r600g driver may become the official way to support older Radeon HD cards and you best way to use that HD6000 you still have laying around in some 2nd or 3rd machine.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]