Slashdot Mirror
3DLabs Releases Linux Drivers
wilfie writes "3DLabs have released linux drivers for their for Wildcat III and Wildcat 4 Graphics accelerators. Being closed source they'll taint your kernel, but what the heck. Press release with penguin-friendly quotes available too." DataSquid has a note about ATI's Linux support: "While on the job hunt, I came across this posting at ATI seeking a project team lead. Last on the list of key responsibilities is "Act as a leader to improve the overall quality of Linux support at ATI." Good news? Certainly better news than what was suggested before."
This is the kind of statement that taints the integrity of the ope source community, and is a prime example of why few commercial companies support Linux.
You will never "find" time for anything. You must "make" it.
Exactly...and just as you have the freedom *not* to use a closed-source driver, you also have the freedom to take advantage of it.
Too often, I see people confusing freedom with politics. Though they sometimes collide, they are *not* the same thing.
Personally, I don't give a crap whether or not the drivers are closed-source. If I ever put the cash down on a 3DLabs card, i'd be a lot more interested in being able to use it on my OS of choice. The freedom to do high-end 3D or video work on Linux as opposed to Windows is a lot more interesting to me than the ability to modify the source code of the drivers.
The only problem is that 3DLabs isn't supporting Linux... They are supporting "linux-somespecificprocessor-somespecificversion"
Might be better than nothing, but not much...
Jeroen
Secure messaging: http://quickmsg.vreeken.net/
3DLabs have released linux drivers for their for Wildcat III and Wildcat 4 Graphics accelerators. Being closed source they'll taint your kernel, but what the heck.
The heck is that we can't port them to other systems and platforms
I am just taken aback that these drivers are not Open Source. The Open Source developer community would have a lot to contribute to these drivers; they could enhance the performance, add new functionality, and make them more robust.
One word: Bullshit. All R100 and R200 Radeon cards have open source drivers. There are, at most, about a dozen people who work on those drivers and the majority of them are paid to do so. Being open source isn't going to make a flock of people go running to improve the drivers.
Dinivin
Yeah, but the drivers they release suck. I bought a Radeon 9000 from them, after the website said supported in Linux. The driver has the option for dual screen, however, there is no way in hell I can get it to work. I have found hundreds of Usenet posts where other people can't get it to work. I haven't found a single post of it working. I emailed ATI on two seperate occasions. The first time I received no response. The second time I asked for a known working XF86Config, and I got a canned response saying they would get back to me. They never did. Right know, I have a $130 piece of silicon that doesn't work. The drivers they do provide only work under XFree86 4.1.0 and 4.2.0. To me, ATI has no Linux support.
-- the computer doesn't want any beer, no matter how much you think it does. NEVER, EVER feed your computer beer.
Unleash the power of your wallet then, by NOT buying their product. They'll either clue in and open their source and start selling product, or they won't and nobody will buy it and their company may suffer.
"I'm a leaf on the wind. Watch how I soar."
-Hoban Washburn
Well, the same reason as to why we don't have room-temp (or only requiring passive cooling) 500mhz processors for $25, silent single platter 10gb HDs for $25, 256mb 266mhz DDR RAM for $25. Flashy new stuff sells, innovation of older products doesn't. Hence why Intel and AMD are pushing up specs instead of improving and lowering the cost of older processors. The HD manufacturers thrive on selling larger and larger HDs instead of coole, more silent and cheaper ones. Hence why we have expensive 250Gb IDE blast furnaces instead of silent 5 to 10 Gb drives whic only cost about 25 bucks. Same thing for memory; pushing up the ammount of memory and speed sells while improving older technologies to be cooler, cheaper and more efficient... Doesn't sell.
Which is kind of stupid really; I'd imagine computers with lower specs but increased stability, efficiency (wasting less power on warming the office) and lower costs would be popular in the corporate scene. Then again, I bet those people are rather thick and convinced by marketing that Office '97 and Windows 98 really do require 200gb of disk space (well, almost) and a P4 3ghz with HT. Not to mention that 512mb of DDR400 and that Ati 9800 that makes Excel run smooth. Woo!
Hate me!
This is a sad offering. They only support specific, official RedHat 7.3 kernel patchlevels -- i.e., there's no compilable kernel module like NVIDIA uses. This somewhat limits the audience for these drivers -- certainly makes them useless for me. I'll stick with my Quadro.
Cantankerous old coot since 1957.
Why? What's the point? It doesn't have 3D capabilities (at least none worth even mentioning at only 8 MB), and 2D video is old hat nowadays. You can get any number of cards with fully accelerated 2D drivers under Linux. The issue is 3D drivers... and then it's only an issue if you want open source drivers that actually perform to the cards capabilities.
I think you have a complete and total lack of understanding about video cards based on your 8MB vs 32MB comment. You realize that 1600x1200 32-bit 2D video uses only 5.7 MB of memory? There are higher resolutions, but they're rarely used. The only need for more memory is texture buffering and z buffering, which are both purely in the realm of 3D graphics. More memory does not have any impact whatsoever on 2D graphics. In fact, most 3D cards are relatively unconcerned about their 2D speeds because it's all "fast enough" nowadays (and yes, I'm old enough to remember when 2D speed was a key measurement -- and remember getting my first card (a Number9 Imagine128, won at Comdex) that could actually scroll text in a window faster than it could full screen).
Taint may well be a technical term but the common usage of the word evokes very negative images, and its usage here is clumsy at best.
Tell someone who doesn't know this technical term that their system is tainted and they'll probably panic, imagining that their PC has been hit by a virus, trojan horse or other undesirable event, where the reality couldn't be further from the truth. (We'll leave the debate about the pros and cons of closed source drivers to another discussion.)
There's got to be a better way of describing a kernel that contains closed source software that isn't so dramatic or apocalyptic. How about "ajar"? At least "ajar" is a better, less ambiguous, description - to me it says "not 100 percent open, and not 100 pecent closed", which is what we're talking about.
"Accept that some days you are the pigeon, and some days you are the statue." - David Brent, Wernham Hogg
That's the point.
By using closed code in your kernel you are putting an unknown in it.
This is bad, the kernel developers can't help you, you can't fix it yourself, you're just stuck with broken software.
I think removing the ability to fix a problem is a dramatic change, particularly when that is a major benefit of free software.
IMHO because it takes *work* to open the source. One nice/bad point of closed-source Windows drivers is that it lets you skimp on documentation and ship just-working code.
"Nice" because every word of that documentation had to be typed and verified, and that calls for a paid employee. "Nice" because the code just has to "work", not be presentably clean.
"Bad" because it means that drivers are coded based on sketchy documentation, informal notes, hallway conversations, and developer memory. Losing a developer may mean having to rediscover proper programming technique for some key feature. "Bad" because "just working" code may not work for all cases, and may be difficult to fix and upgrade for the next chipset.
So in the long run, "Nice" may be cheaper than "Bad" or at the very least not much more expensive. But in the short run, "Nice" is always quite a bit more expensive. Few in the computing industry manage to have a long-range view.
As for the IP arguments, IMHO the only valid ones are contractual. Drivers can be disassembled, and I've been on both ends of silicon delayering and analysis. It's not like you're trying to rebuild a clean-room clone - there are targeted features you go after, not a schematic of the whole design or source of the whole driver.
The living have better things to do than to continue hating the dead.
My impression on at least one of the reasons why they do this, is that there is certain proprietary code licensed from other companies in these drivers, which they simply do not have permission to release with an open license. I suppose the primary manufacturer (ie, nvidia, ati, or 3d labs) could work to get the people who do own the proprietary licenses on that code to open it. However, this seems like a lot of work from a video card chip manufacturer's perspective, for (debateably) little gain. I've also heard grumblings (maybe just rumors though) that companies are afraid to release their drivers open source, because then competitors can see how their architecture is set up.