Slashdot Mirror
XGI, VIA Release Open Source Drivers
An anonymous reader writes "XGI has announced the release of open source drivers for its Volari family of graphics adapters. Efforts at X.Org to merge the new code into the head branch are already underway. Almost simultaneously, VIA has announced the immediate release of open source drivers for S3 Graphics UniChrome, VIA ProSavage and ProSavage DDR. Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards?"
So based on this news what is the best card to buy?
It isn't like they have much to lose. They sell hardware, the drivers simply let that hardware operate. They probably also know this will earn them points with the Open Source Community.. which is always a good thing.
Proceed with Format (Y/N)? Y
This is good news for fanless C3-based systems using CLE266 MPEG acceleration hardware: Via had released closed-source drivers (and, indeed, forked Xine to use them in a product called VeXP). These drivers were reverse-engineered to support an open source equivalent, which was less than completely reliable.
You could've hired me.
You paying attention to this ATI?
And guess what? I think this will ultimately increase their sales. As more drivers are available, choosing hardware to run Linux or some obscure OS won't be so difficult, so people will be more likely to buy a piece of hardware.
I think this will also improve the quality of their products. Often, drivers, like any other software, contain bugs, which can cause it to appear as if the hardware isn't working as well as it should. Or perhaps the driver isn't quite as efficient as it could be with system resources, so it seems as if the hardware isn't quite as fast as it should be. When these things are released under open source, it is more likely that things like this will get fixed and improved, and that will ultimately improve the vendor's hardware product without requiring any significant effort on the part of the vendor.
XGI and VIA are doing a smart thing. I'm heading over to write them an email about them and thank them. I suggest that others do the same. This is great news, and I hope other vendors will follow.
When looking through the kernel source code there is only support for 2D. Kernel bugreport X.org bugreport
...or you could look at it as giving people only what they are paying for.
Is it really a racket, then or is it actually more of a manufacturing strategy? On the surface, we all want to think that price should be based on what it costs to make it. But there's more to it and, really, the only time material, labor and overhead costs come in to play when pricing an item is finding out where your break-even point is. After that, it's essentially demand setting the price. The fact is, that there are several markets... the home user market, the professional user market and on and on.
And forget that you're a technical type and think like a business man who doesn't know tech. When you are told you have these three graphics cards to choose from, each with comparable capabilities, one of them has this ridiculously low price for its class. Are you inclined to buy that one? Most business people don't because it causes them to doubt the lower-cost device. "Why is it cheaper? Must not be as good."
But back to manufacturing, it's important to lower manufacturing costs where ever possible... if it were your job to do it, you'd probably be no different. It's cheaper to make a bunch of the same product and then disable features and sell them as lower-end rather than to manage that many more product manufacturing lines.
Is it frustrating to the technical consumer to know this? Hell yeah. I've got a Dell Inspiron advanced port replicator and a Dell Latitude advanced port replicator that are freaking identical hardware and they work interchangeably except that some ports don't function properly. I haven't decided to crack these two things open to find out what's different, but there is a fairly significant cost difference between the two devices.
Is it a racket? No... I think that goes a little too far.
The XGI release is 2d only (the kernel code is for fb support, not DRI), and from what I see on the Unichrome driver effort's mailing list archives the VIA source release is just making available to everyone what has been available through a "developer portal" for some time and does not make any more of the chipsets' features usable.
So the only possible real news here is a shift in the attitudes of these companies. We'll see how that works out in the future (whether enough information is released to allow open-source 3d drivers for XGI and full support for the VIA MPEG enc/dec acceleration).
Cool. XGI took a very old version of the SiS driver, stripped out many useful features, uglyfied the code beyond belief and calls this *their* effort of open source development.
And the worst part is that my name is all across "their" source.
Finally, probably needless to say, the 3D part is not included.
-- The Online Photo Editor - http://www.phixr.com
This is not completely fair. Most hardware companies depend on code in their drivers that their staff did not write. As contracts generally go, the outside developer usually imposes limits on use and distribution of their work. It's invariably more expensive to purchase outsourced code without restrictions.
$5 / month hosted VPS on linux = awesome!
Won't compile with 2.6.11? Check.
Compiles with BIG, LARGE warnings about depreciated features being used in 2.6.10? Check.
Won't work under x64_64? Check.
2D part of drivers buggy? Check.
Infrequent releases that don't correct problems? Check.
No support for X RandR? Check.
Sorry, the ATI drivers don't pass muster. Perhaps I should've realized sooner with the constant weird 2D bugs I had with the ATI driver. Or the fact it wouldn't compile on 2.6.11. Or the fact it just plain won't work as advertised on 64-bit Linux.
I took out my Radeon 8500, put in a Geforce 2MX I had, and installed the nVidia driver. It was actually wrapped in an installer, rather than me having to manually untar and run scripts ala ATI. It asked if I wanted 32-bit compatibility OpenGL libraries. It told me that the 2.6.11 kernel fixed some AGP issues and was reccomended (which was good since I already had it, and only used the 2.6.10 because of ATI). X RandR started to work with the nVidia driver. 64-bit and 32-bit apps work flawlessly with each other.
ATI is shit. Their card hardware may be good, but without a driver, it might as well be an ISA SB16 for all the use I get out of it.
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.