The Letter That Started AMD's Open-Source Strategy
An anonymous reader writes "In marking the fourth anniversary of AMD's open-source strategy for their Radeon graphics, Phoronix has published the letter that launched this open-source effort. It was a letter written by Novell SUSE X engineers and submitted to AMD management with their open-source proposal."
Even their closed-source driver sucks, it's not just the open source one.
The OSS driver works pretty good for antique hardware. Unfortunately, it doesn't work very well for anything vaguely modern, while fglrx pretty much doesn't support anything more than a few years old (and it does more or less suck.) Consequently, if you have anything but the fanciest (unless it's very very new) or shabbiest ATI card, you can expect it to suck rocks through straws on Linux. nVidia is better but shares many of the same flaws. However, middle-aged hardware is well-supported by the official driver, and amazingly old hardware is supported as well. That makes support much easier, and while shopping for older computers with Linux compatibility in mind, it makes avoiding ATI a no-brainer as well. This reduction in resale value causes me to value ATI less up front... But to the masses who will never run Linux on a desktop, it's fairly irrelevant. Most people don't buy used hardware.
Anyone want to buy a P4 desktop with an ATI Rage Pro in it? It runs Ubuntu just fine :)
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
...a non-free firmware blob is required to use those drivers. Can we really call them free?
All I see is an article talking about the letter, where's the actual letter?
--
Stay tuned for some shock and awe coming right up after this messages!
The first few posts might be trollish but look at the original article. It's a puff-piece wrapping AMD in the open source flag as a paragon of virtue when the truth is far less rosy.
AMD enjoys a (largely unearned) reputation on Slashdot as being the scrappy underdogs who want to save the world for Open Source or something like that. People forget that AMD is (or was) a multi-billion company that has no problems with getting patents, cozying up with Microsoft (see Jerry Sanders testifying in favour of MS at the anti-trust trial... ironic), or locking up its IP in the video card area. Sure they dump partial sets of documentation several months after releasing a new card, but with that level of support I have no reason to upgrade from a 3 year old Nvidia card since the shiny new AMD card won't work any better.
I *want* AMD to have real open source drivers that give great support for their hardware... basically what Intel already does right now. However, congratulating AMD and holding them up as superior to everyone else when they are not is not the right way to accomplish this goal.
AntiFA: An abbreviation for Anti First Amendment.
I believe there was also support for opening the then acquired ATI wares from inside AMD. A former KGI/GGI Project developer (along with fellow AMD employees) were pushing for AMD to open the specifications to ATI's hardware.
The difference there is that Intel doesn't have any IP in this area worth protecting.
Also, what you're conveniently ignoring is that most of the source that AMD has came from ATI and was prior to the change in strategy. It's not easy to go back and retroactively open source things for which you may or may not already have the rights. I'm guessing that there's probably a fair amount of other people's IP involved. And even if there isn't, the legal team does still need to go through and make sure that they aren't going to be sued for releaseing something they shouldn't.
To ensure an open development process NOVELL would require that it will not make use of any specifications or programming documentation that can not be made available to other developers from the open source community also under a suitable documentation publication program which will permit the release of source code under an open source license.
This step will help to ensure continued maintenance for hardware components beyond the maintenance cycle of the manufacturer and will help customer to secure their investment. Furthermore it will demonstrate and underline AMD's full commitment to the open source development model and send a positive signal to the open source community which this has been waiting for for a long time.
NOVELL will ensure that a driver with at least base functionalities is available for earlier releases of the X Window System at least back to X11 R6.9 to be integrated in existing enterprise products by their respective vendors.
Well it works fine as a marketing lie apparently. I recently had the joy of upgrading my home PC, and there was hundreds of mid-range graphfcs cards to choose from. I got the AMD mainly because they support open source (or so I thought), and I may want to switch to Linux some time in the future if they can finally get the video rendering up to the same level as Windows.. (I have used laptops with AMD and NVidia graphics card on Linux and I haven't noticed much difference actually. The most recent one was an NVidia card, and I could never get rid of the tearing artifacts on the secondary display)
lspci | grep -i radeon && exit 1
done.
I beg to differ: I've run R600 and R700 (read: 3000 and 4000 series) cards with the FOSS driver and I find it miles better than fglrx: at least it does play nicely with all other components in the system.
A CC-licensed illustrated horror novel
I sure wish my X1250 hardware would work. I have a subnotebook on which I can only run Vista because of drivers. It has R690M chipset and L110 CPU and it's generally poorly supported under linux, the power scaling stuff doesn't work right either.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
The "what have you done for me lately" crowd has an interesting way with words that never fails to amuse me.
Everyone else? There are three horses in this race, and a few ticks hitched to the feathers surrounding the coffin bone. 90% of the cells in the human body are bacterial. Sometimes you have to integrate over quorum.
Intel isn't especially huffed about offloading high-demand computing to a GPU chip manufactured by somebody else. Without somebody else poised to siphon an artery, Intel would do everything in their power to level the field. We would have one coefficient of Moore's law governing performance, rather than two. Generally when you lash two horses together, the slower horse governs the pace.
Fortunately, the GPU swallowed the blue pill before Intel could do much about it. On the side of the fence with the big Cheshire grin, we have precisely two spectral lines of any significance: red and green. But you have to remember that both of these companies exist in an ecosystem where the bully in the china shop hoovers up the vast majority of the resources like the human race arrived on a shiny new continent.
It ain't easy feeding the sourdough culture known as Fab. It's the pudding mix in Sleeper, the plant in Little Shop of Horrors.
If your Fab blows a bubble, you're in a world of hurt. Around this nightmare, you have to knock off some of the most technically demanding design projects known to man, year in and year out. After you build a Saturn V, everyone wants a Saturn VI. The Saturn VI blows everyone away--for a year or two--then everyone starts to itch and scratch for a Saturn VII.
Back when the original Saturn V was crawling toward heaven at a top speed of 1mph, who exactly was "everyone else"? But let's not give the Americans any credit for trumping anyone else.
The painful truth here is that if you love open source, sometimes you have to settle for second best. AMD is slowly making good on the promise of taking this track, although it's truly frightened to think of how much oxygen has boiled off in their seemingly perpetual state of launch readiness. It ain't called Fusion for nothing.
At the bottom of the process this is an IP issue. Some people seem to think that open source happens in Wikileak time frames. It could work that way, but billion transistor designs would really stress out your onion router, and customs might seize your next mask set.
I played a chess game not long ago where I settled into the Siberian Winter defense. In other words, my opponent was better than me, but he fired his powder a moment too soon.
I was boxed in by the mate threat, and his mate threat was boxed in by my passed wing pawn on the other side of the board. If his quill armada didn't crush me first, my winter pawn would crush him later.
If you're asking "what has AMD's wing pawn done for you lately" I suppose the answer is that it sits there doing not very much.
In my case, not very much won me the game after 10 rounds of desperately accurate counter-parries.
Three cheers for the winter pawn.
There are some hardware solutions that suck so bad you wouldn't want Linux on their anyways.
- Dan.
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
DON'T BUY FROM ATI enemy of your freedom
That picture is looking sillier and sillier as time goes on.
my amd fusion e-350 begs to differ on that. it works very well on the foss drivers.
obvious troll is obvious.
In capabilities the graphics are worlds ahead of anything from intel. And the CPU is also pissed off. It works great under Vista. I get 4:30 battery life (no, really) and everything works, I can play games, whatever. Under Windows 7 I get 2:30 battery life, and many games blow up the driver. Under Linux I get 1:30 battery life and even with RenderAccel, GLX etc disabled I get trashing of the display whether I use vesa or ati.
AMD royally screwed the pooch on Linux support for R690M and L110 and I am forced to run Vista on my portable as a result.
Yes, I should have done more research. This is what I get for trusting AMD to provide Linux support. Never again.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
ATI has great quality hardware, but lower quality drivers/software. Nvidia it is the opposite.
I switched from Nvidia to ATI for this reason. My ATI 5750 does run Beryl suprising well. I game only on Windows 7 and I doubt performance would be good in Linux, but that can change. If ATI could get good quality drivers for Linux we would be happy to support them. The specs and code open are cryptic and only cover what appears to be a dispatcher which then transmit the code to the different parts of the GPU according to other posters here.
It is not like we are going to compete agaisnt ATI with trade secrets unless one of us has a 1 billion dollar chip fab plant.
Intel opened their speced and it helped them tremendously. Now since ATI has great integration with their bulldozer and Llamo chips new innovations would help sales. We could even improve the drivers to the point where some of the code can be contributed back to their team who develops drivers for Windows.
I have noticed that World of Warcraft runs slower in DirectX 11 vs DirectX 9 which is odd and points to the drivers needing work as Nvidia users get a 20% performance increase. Opening will help.
http://saveie6.com/
I do. Or rather, I did...
http://www.geeks.com/details.asp?invtid=SAMBA845V-24-4-R&cat=SYS
At that price, it's a great deal. Start a kickstart install of CentOS5.x, and deliver it to customers who need a bunch of office desktops, terminals, etc. These days, you can spend a lot of money to get super-efficient components, and still end up drawing more power, and making much more noise (above PC idles at under 40watts, and is impressively damn-near silent).
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
Question: Have you tried SIW to find out what the exact chips you are running that are having Win 7 driver problems, and then replacing them with an updated driver from another model with the same chip?
Because I have found a LOT of the time when it comes to laptops when the OEM says "Windows X not supported" it means "We didn't bother with a driver for it" and I bet my last nickel that there is NO part on your machines that SOME OEM didn't write a better Win 7 driver for than what you have now.
Here is how you fix it, find out the name of your device that is having problems, lets say it is the R690m. Then all you do is Google "R690m Windows 7 driver" and choose the latest one you can find. Then simply extract the driver from the .exe (both winRAR and 7-Zip can do this) and then as I put it "Rub windows nose in it" by installing the driver manually which if you are on /. I'm sure you know how to do a manual driver install.
That's it! In less than 30 minutes you'll have a much better windows 7 driver and all those boo boos go bye bye. you would be surprised how often I have to do that trick, especially with wireless. just last week I had an HP with Vista only drivers and when the guy tried to install 7 himself it hosed his wireless. I just used SIW to find out it was a Ralink chip, found a Ralink driver for win 7, rubbed windows nose in it, and there you go. One more happy customer.
You just have to remember that with Windows there is ALWAYS more than one way to skin a cat. sometimes it is the most simple way, sometimes you need to add a step or two, but you should always be able to find a way around the boo boo. Hell I have a granny that bought the Win 7 HP 3 pack and decided to have me load the last license on "her spare" which was a 2.2Ghz P4 that was old as dirt. took a little tweaking but now she is happily running on it while I have her new machine here in the shop. somehow she managed to completely wipe ALL the fonts off the machine, ugh. I swear that woman could kill a Sherman tank with a toothbrush.
ACs don't waste your time replying, your posts are never seen by me.
Question: Have you tried SIW to find out what the exact chips you are running that are having Win 7 driver problems, and then replacing them with an updated driver from another model with the same chip?
Yep. I searched long and hard to find a driver that would work, and finally did. On the second resume from suspend, I get a free reboot. There is no improved power saving driver, either.
You just have to remember that with Windows there is ALWAYS more than one way to skin a cat.
The cat is a lie.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
The tearing on the secondary display is present in every card. I've had it since my first GeForce2 MX 10 years ago, GF4MX, GF 5200, 7300GT, 6150 IGP, intel GMA950 (i think), and now with Intel Core i5 on-CPU graphics. Seen it with VGA, DVI, Component, and HDMI outs. The secondary display always has tearing. That's why I have to switch primary/secondary screens when I want to watch a movie on my TV.
Not sure how it is in AMD land, but on nVidia and Intel, that's what it's like.
Name the part and I'll be happy to find the driver FOR you. Just let me know which one you tried so I don't duplicate the effort. You can then get a copy of "Windows 7 Tiny" to test it, which only uses 3Gb of HDD space installed.
Hell did you just try using the Vista driver? because the Vista/7 driver model is damned near identical, just as you could run Win2K/XP/2K3 drivers interchangeably unless you were switching from 32bit to 64bit. I personally have Vista X64 drivers running on my Windows 7 install simply because nobody bothered to write a Windows 7 driver for my USB capture card. Works beautifully BTW, great picture and integrates perfectly with WMC.
BTW did you notice I got modded down for daring to try to help you? How sad that the Moonies here.....err...I mean FOSSies, have such a groupthink circle jerk that the one and ONLY posting allowed within a thousand yards of a FOSSie story is "Gee Biff, isn't Linux perfect? Why it sure is Skip, and RMS's farts smell like roses!"
Sadly as you found out trying to run it on a laptop and I as a retailer found out trying to actually sell Linux it is MORE WORSE NOW THAN EVER thanks to Torvalds constant kernel fiddling and the driver mess. Out of the 8 machines I picked at random to test Linux on, tried a half dozen distros BTW, guess how many survived with full functionality after first upgrade? NONE. Zip, zero, nada, squat. Wireless, sound, video, Ethernet, there was ALWAYS something broken.
I could provide links to fellow retailers running away from Linux, or point out that Dell has to run their own repo just to keep drivers from breaking, but what's the point? Just like Moonies they'll just stick their fingers in their ears and go "CLI is leet! copypasta into term makes our neckbeard thick! You must be M$ Ninja!" and go back to their circle jerk, so why bother. let them enjoy 1% and ignore their crazy asses I say.
ACs don't waste your time replying, your posts are never seen by me.
in AMD land you get some minor screen tearing but on all displays. At least that is my experience with a 4200HD (on board) and a dedicated 6850HD.
It isn't a showstopper though and thus I haven't tried to fix it.
The above goes for all driver editions radeon, fglrx, mesa-dri-[a-z]{3,8}-exp.*
-- no sig today
You probably got modded down for insisting there is a solution. Of course I tried the Vista driver. I tried over a half-dozen drivers. Just finding one willing to install was a bit of a nightmare. And I must add that I never did get a driver to charge my Motorola cellphone even, and I tried like ten of those that said they would work. Windows 7 is a bit of a driver nightmare.
When I tell you that the CPU is Athlon L110 and the chipset is R690M I have told you literally everything you need to know, because that's the whole fucking computer and it's the CPU and Video I'm having problems with. But if you really want to know in detail what it is, it's a Gateway LT3103u. You should be able to find a lspci if you look around.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
They're not trying to keep average Joe out of the GPU manuf. industry, they're trying to keep Chinese companies who already manuf. chips at TSMC from having the code to the drivers. The chinese already have the complete hardware design of the chip, but its worthless without the huge investment in driver design because the two go together like peas and carrots. Good closed source drivers may be one of the few things that prevents rich chinese companies from entirely taking over the GPU market.