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
I wish all other hardware companies did this.
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?
These companies have absolutely nothing to lose by releasing their driver "secrets" into the open. They are both non-players in the graphics card market.
I doubt this move will have any influence on ATI or Nvidia to open things up but we can always hope...
The quality of these drivers becomes extremely good in the X tree such that both ATI & Nvidia lose sales because of it.
The only way those two will release their own drivers as open-source is when they feel a pinch in the pocketbook.
...to the question about free drivers: yes, maybe, I certainly hope, all of the above.
We already get our drivers for $free on most platforms. Free as in open and easy, I certainly hope so. While we've made great strides in compatibility over the years, getting the hardware people on board and co-operating is still lagging.
When 95% of each sector of the hardware market is co-operating, then we'll just have to hope coders are doing something useful with the platforms now that they're working.
I'll put this in my personal "Good News" category for future reference.
If my grammar and spelling are off, I am [distracted/tired/careless] (take your pick)
"Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards?"
I doubt it. Just a coincidence. Wishful thinking. Once nVidia releases open source drivers, you may start to think otherwise.
Digital Sailor
Maybe when ATI and nVIDIA release open source drivers as they are the drivers (pun not intended) of graphics technology
Free MacMini
If the performance on those cards is anywhere near decent I'll be buying one.
I don't know the meaning of the word 'don't' - J
I've heard that newer NVidia cards can boot straight to TV.
Now I just have to decide on whether I wait for someone to work out a Open source driver for the XGI card or just spend the ~$40 on a NVidia card when I have a perfectly decent XGI card already.
heh- who am I kidding. I'm cheap. And patient.
Come on guys- let's start reverse engineering these XGI drivers!
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.
Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards?
Only if nVidia and/or ATI follow suit. (I know that in some cases they can't, but they could take an approach like Netscape and Sun did, release everything you do own and leave out the stuff you don't).
When looking through the kernel source code there is only support for 2D. Kernel bugreport X.org bugreport
Volari, oh oh, cantari, oh oh oh oh
Let's fly way up to the clouds, away from the maddening crowds....
Are we getting a fully open source GLX driver as well? Does that GLX driver exploit all the features of the chip? If there is hardware shaders can the GLX driver use those? Or are we getting something like the open source "nv" drive that only does 2D?
Unstable Apps: Our Android Apps Don't Suck
Even though there are open source video drivers for All In One Radeon, there is very slow development on All In One Radeon tuner applications. I have gone to gatos.sourceforge.net and downloaded the tuner drivers for an older X.org, but now that X.org is improving, I can't watch TV on the latest X.org. I have trouble compiling from source after half a dozen attempts
"True, but a lot of hardware companies hoard their IP as though it has intrinsic value. They seem to assume that since it cost something to produce, it must be worth something."
One way to find out. Give it to your competitors.
"Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards" VIA is hardly a leader in the video industry. Seems to me like this type of prediction is just going to wind up letting people down. Hope is nice, but lets not slide down this slippery slope.
Whatever the background, it's wonderful news. Maybe nVidia and ATI will see some real competition now from these manufacturers with open-source drivers, and push to find a way through the 3rd-party licensing morass which supposedly prevents them from open-sourcing their drivers at the moment.
Does anyone here consider the head to be a branch? IMHO a branch is taken from the head. The head is just a trunk. Not a branch.
But my video drivers are already provided free with the box... why duplicate the effort?
Firmware allows the same card to be resold as a 'professional workstation card' or as a 'consumer card'. I can't remember exactly which but I think some of ATI's FireGL's are identical to several of their 9x00 or X800 cards.
Secret drivers no doubt also contribute to this scenario. It's just a way of conning people into different rebranded (but physically identical) products.
"The quality of these drivers becomes extremely good in the X tree such that both ATI & Nvidia lose sales because of it."
If wishes were fishes. So what's the open source timetable for video drivers as good as what Nvidia's producing? Month? Two months? Year? Two years? Keeping up?
"The only way those two will release their own drivers as open-source is when they feel a pinch in the pocketbook."
A threat is useless without teeth. That reminds me. How's that open source killer video card coming along?
+5 Dean Martin
This is good news. I will be giving XGI and VIA priority on my next hardware purchase.
Got burned by Matrox (triple head card) when they dropped Linux support after kernel 2.6 came out. They now point you to a third party where you can pay ~US$200 for what Matrox provides to windows users for free.
Currently using Nvidia. Not open but it works and is easy to install.
Non-players soon to be players.
But they paid you to post what they did, didn't they?
Huh?
I was able to boot directly to TV (svideo) with my nVidia GeForce 2 MX in linux no problem, I don't think that feature is only reserved for newer cards.
I'm interested in seeing how XGI will do with full opensource support. I'd think the opensouce community should be able to help XGI optimize it's drivers. Sure ATI/NVIDIA will still rule the Windoze gaming market, but I can see a cult following developing behind XGI real soon.
Man is the lowest-cost, 150-pound, nonlinear, all-purpose computer system which can be mass-produced by unskilled labor.
"It was mentioned in a comment on slashdot somewhere that the code in the ATI or nVidia drivers may be propriatary and closed source as liscensed from someone else. They may have bought code from company x but company x may not allow for that code to be open source. So instead of re-writing the drivers entirely so that nVidia/ATI own all of their own code, they may just stick with the binary drivers to protect other companies IP."
And of course we ONLY have their word that that's the real reason.* Sounds like the same reason they used for the nForce ethernet driver...until someone reverse-engineered it. Then they suddenly became all helpful.
*Security (from us) through obscurity.
I'm glad to see this positive development from XGI. Releasing open source Linux drivers can't help but give them positive publicity. Back in November of 2003 I said we had "an opportunity to persuade [XGI] that supporting Linux by releasing drivers would gain them positive reviews and have an impact on sales." XGI has released the drivers, now it remains to be seen whether this drives sales.
I don't know who persuaded XGI to make this commitment to open source but I fully intend to consider XGI for my next video card. I'm using an nVidia card today on the basis of their closed-source driver support for Linux, but I'd rather support a company that embraces open source.
This is simply a move back to how it was in the old days (80's / early 90's) of PC's when lots of lowlevel technical information was provided with graphics cards. Good to see this happening (again!)
Read up on video cards and Macrovision ... if companies making popular cards like ATI and NVidia start open sourcing drivers, it'll allow end users to trivially circumvent DRM plans (also read up on the whole 'trusted path' idea ... encrypted straight from CPU to monitor).
It's time to shift your support to companies that support open source (and by consequence, oppose DRM).
Is there a PCI-e card that I can buy to replace my OEM ATI card that'll use these free drivers? The card is pretty new, 128Megs and such, but slow, so if I could buy an upgrade *and* get away from the ATI drivers, I will...
bo
bad_outlook
--
Is this vague enough for you?
The XGI drivers are 2D-only; you still have to use the binary library provided by XGI for 3D.
This sucks, too, because the performance of the XGI Volari V8 is comparable to a Radeon 9600 or Geforce 5700. And I'm sure that their drivers suck, so there's probably more performance in them. And it's dirt cheap, too. A 256MB card comes in at just under $100, and a 128MB card at $85.
XGI needs to be told that this isn't enough.
Why? Well, I have a bunch of machines here that use the KM 266 Via Chipset and support is scarce. Even proper 2D support is rare (I've only seen it in Fedora 3 and SUSE). Other than that, most other distros will either report 24 bit color (it's 32), or force you to use the generic VGA driver (which is SLOWwww).
And 3D support? Non-existant. Not that the 3D is spectacular on the KM series anyway, but it's certainly passable for screen savers, programs like Celestia, and other non-'Doom 3' purposs.
And it's not as though the KM 266 isn't capable of better. Under Windows it performs just great for what most of my users want - just not under Linux.
Thank you VIA, it will only help you...
"...Well, there's egg and bacon; egg sausage and bacon; egg and spam; egg bacon and spam; egg bacon sausage and spam..."
I haven't managed to get the DRI drivers to compile, install, and run from CVS/SVN in quite some time - I keep hoping there'll be a new release that supports it (yeah, I know, it's not ATI or NVidia performance, but I can't exactly swap the graphics chip in my laptop...)
I tried following the links, but the download section for my actual ProSavage/DDR chipset only has Windows drivers...
Hacker Public Radio is our Friend
Well we seem to have slashdotted XGI at http://www.xgitech.com/
:)
So I'm sure they are getting good feedback
Now if their sales increase at the same rate...
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).
who here has actually used an xgi card, and what was your experience?
I am extremely curious--they are cheap, but I want to know about performance.
"We don't know what we are doing, but we are doing it very carefully,..." Wherry, R.J. Personnel Psychology (1995)
I ask because outside nVidia, Windows drivers are usually lacking one way or another, specially when it comes to stuff like OpenGL. It could be interesting if someone worked independently on Win drivers for mainstream videocards.
PS: It's great to see some companies realizing they are on the hardware buissnes, not software. Thank you. I had interest in the S3 Deltachrome/Unichrome series, and now i'll most certainly try one out.
I applaud the efforts of these hardware vendors to support Free drivers, but:
These guys are NOT the market leaders - they are very much in a niche market. They are behind and falling even more behind in terms of market penetration - as more embedded systems move to Intel, nVidia, and ATI these guys lose ground.
They are hoping that by having Free drivers, they will pick up some sales in embedded widgets that they might otherwise not get. Yes, these guys make their money on hardware, AND they don't feel there is any "magic" IPR that can be exposed by releasing their programming specifications.
So, what is the likely impact upon ATI and nVidia?
Almost none.
Now, *IF* we start seeing ATI and nVidia slipping in their sales to hardware vendors (laptop makers, makers of embedded widgets, etc.) *then* we *might* see ATI and nVidia getting worried. And if a winning lottery ticket happens to blow into my hand I might be a millionare.
But since the segment of the embedded market to which ATI and nVidia sell (as well as the laptop market and motherboard manufacturing market) are still 90% or more "powered by Windows®", there is still little motivation for ATI or nVidia to re-examine their stance on releasing information about their chips - they will continue to fear that releasing programming specs will somehow reveil the "magic" in their chips.
www.eFax.com are spammers
So, in return for them being one of the first gfx card makers to release OSS driver, possibly starting an OSS revolution in the graphics card/hardware industry, we melt their web server?
95% of all computer errors occur between chair and keyboard (TM)
...I can play Unreal 2k4 and Doom 3 with my ATI Radeon All-in-wonder. It's not X.org's fault though. It's ATI's fault because they are idiot arrogant bastards who won't release specs. I have two of those cards and they are sitting in boxes because there is no really good X driver that supports 3D acceleration. I've had to go over to the dark side and buy NVidia cards (they are the dark side because they killed off 3DFX who WAS Linux friendly). That's my only option though. Either that or just don't play the games. As it is right now, I only have one Windows box at home and it's probaby going to disappear in about a year. (Thanks to the Fedora crew!)
-"...bad old ideas look confusingly fresh when they are packaged as technology" - Jaron Lanier (Digital Maoism on Edge.o
Unichrome has basic 3D hardware and the new drivers seem to support this.
Slashdot needs a new moderation type - "Potentially Insightful Answers". This moderation class should +1 a post, however whenever any child post is given +1 Insightful, the parent should have 1 point removed (to a minimum of 1). So if someone asks "Where in Soviet Russia can I find Natalie Portman's hot grits?", they can be moderated up to give visibility to potential answeres. However once someone answers the question, they are moderated up at the cost of the original post.
As it is, it's a tried-and-true karma whoring technique of asking an obvious, every-wants-to-know question as quickly as possible. Every moderator that wants to know the answer will mod it up, hoping that truly insightful/informative people will come along and see it. Yet the question itself isn't insightful/informative whatsoever, and is usually a very cheap post.
Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards?
/. story here). for those who don't read the Developers section.
There's even open source hardware from the Open Source Project (OGP) coming out (info here and here, and the
The PCI version is due soon, and reported to have resolutions up to 2048x2048, dual-link DVI and TV-out (but won't be capable of playing HalfLife2 or anything like that).
SEO Copywriter. Just Say ON
You actually put the difference between their/there/they're in your sig, but you get loose/lose wrong? Come on...
I would also be interested in a good, cheap, silent, low power 2d card for my HTPC, which card did you excactly have ? How much for it ?
For the boot to tv, i think you can try to fool the card that there's an monitor attached, if it really demands a connected monitor. (just need to short some pins at the vga connector)
I just downloaded their drivers and they seem to include everything you need for 2D, 3D and mpeg2 decoding.
kernel 2.6 drivers
It's nice to see a company actually releasing useful open source drivers.
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
We have noticed this development and we are taking notes in the event we become a second-rate video card vendor with pitiful marketshare. Until then, we'll be continuing to hoard our driver secrets so we can eventually kill nVidia and become the video card overloard. Pay no attention to the quality of our drivers compared to nVidia.
Sincerely, ATi
ps: When we become overlords, please do not continue to ask for free drivers, as we can not afford any insurgency in our dominion.
The cards everyone really cares about, the nVidia and ATI ones, are covered in several layers of NDA. Hell, we don't even know what ISA the GeForce 6 uses, and AFAIK the Radeon X--- series is similarly shielded.
The chipsets in question are minimal-functionality low-performance cards which can't even remotely compete with top-of-the-line chipsets.
:(
Linux support for low-end chipsets has never been lacking. This release isn't news at all.
If you want decent performance and reliability, NVidia's binary-only drivers are still your only option.
retrorocket.o not found, launch anyway?
whichever went first would score a real coup against the other.
You guys can't even imagine how this IS important for wilde scale low-end computers out there. Binary drivers from VIA have been always crap, and let's hope in the right hands they will make Linux desktop expierence for lot of people much much better.
:)
p.s. Yeah, it is God with big G. Don't ask me
user@ubuntubox:~$ stfu This server is going down for shutdown NOW!
I certainly welcome more open-source drivers, but the reason there aren't as many isn't just because companies think they're in the business of producing software drivers for Microsoft. They also fear that opening up their drivers will give information to their competitors in the hardware market.
With graphics cards in particular, there's actually quite a lot going on on the driver level, and some of the "trade secrets" are even how to write optimal drivers themselves---if nVidia's drivers are 10% more efficient than ATI's in utilizing the card, it makes their card look 10% more efficient, and they sell more of them. You can also glean a lot of details about the hardware if you have the driver source code, especially the original un-obfuscated source code that lets you infer what the programmers were thinking when they wrote it.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
"Could these moves signal the beginning of a period of rapid improvement in Free drivers for video cards?"
:) (I saw it played *very* smoothly even 4 or 5 years ago on Dell laptops with built-in 3D acceleration, but I always buy low on the computer food chain. Which means I can't really complain, per se, but I can still hope.)
Boy, I hope so! Most Linux distros nowadays are really nice out of the box, graphically -- for 2D. Since I am close to a non-gamer (weaknesses include frozen bubble, kbounce, and similar addicitive brain-rotters), this isn't much of a problem for me.
(And Yes -- you can disagree with me happily, but my main issue, as a distro-fickle klutz and permanent newbie, is how it behaves out of the box. If I have to download, compile, install, change settings, perform voodoo, my incentive must be that much greater.)
Some cards obviously work better than others, but in the machine I'm working on right now, I happen to *have* built-in "S3 Unichrome" graphics, and it would sure be nice if I could finally play TuxRacer smoothly
timothy
jrnl: http://tinyurl.com/c2l8yr / foes: http://tinyurl.com/ckjno5
Both nVidia and ATI have much to lose if their driver code gets out. Their specific hardware and software optimizations, relatively easily gleaned from the driver source, would be as gold in their competition's hands. Also, they've both been caught cheating by optimizing for specific benchmarks/games to boost their framerates in order to boost their bottom lines. When the difference in selling your new top-of-the-line video card for $500 US versus being in second place and your top card only able to command ~$350 or so, there's a lot of money on the line. And with the six month product cycle that we've seen for the last few years, there's a lot of money to be made or lost.
No, the opposite is true - at least for SiS cards (which are pretty close to XGI's): If there is a monitor connected, they will NOT enable s-video. Physically disconnect the monitor before booting, and have the TV switched on.
-- The Online Photo Editor - http://www.phixr.com
With all of the recent discussions of reverse engineering BitKeeper, I wonder if it is possible to reverse engineer the ATI and nVidia drivers. In the same manner that the first IBM PCs were done (ala clean room.)
/*
* Copyright 1998-2005 VIA Technologies, Inc. All Rights Reserved.
* Copyright 2001-2005 S3 Graphics, Inc. All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sub license,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
* VIA, S3 GRAPHICS, AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
I was able to change the primary display on my GF4Ti4200 (is that newer?) to the S-Video, and have it use the S-Video exclusively, including the boot, POST, et cetera.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Thanks to the hard work of Tim Roberts (and lately Alex Deucher) among others who have made it possible to run these cards for many years now. Check out the main site for Free Savage drivers
No, really, how can i fbe used against them? There's nothing about what it is doing internally, so you can't use it to engineer a workalike GPU.
What?!?!
If only Nvidia could do likewise. Although they've done some wonderful work supporting their Linux drivers, a quick browse of the nvidia linux forums (which are by far the most popular at the nV News forums - not coincidentaly) shows the huge number of people who've been experiencing bugs (and in particularly the "screen frozen, but mouse pointer moves" bug) who would love to help, but can't.
Graphics card drivers are precisely the software that needs to be open source. It is simply too crucial an element of a stable system to let one company attempt to handle the vast range of systems and configurations out there.
Well at least the BSD license flavor is alive and well.
"I've heard that newer NVidia cards can boot straight to TV."
They sure can. I've done that more than once, in scenarios where I had no monitor at all save for a TV screen. In fact I'm actually surprised there are cards that can't do this.
I have the Volari V3XT (I believe). I really would rather just pull the monitor over as opposed to shorting some pins. I also got the V3XT because it has DVI out- and someday I'll want a new TV.
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.
Oh, it's a lovely idea. But XGI's shifty and deceptive hijinks pretty much bork the idea of a complete switchover, and nice as VIA's integrated hardware is, it doesn't hold a candle to any reasonably modern solution by the two main graphics card companies. Now I fully support VIA's decision here. It's been a sticking point for lots of people, and is very much The Right Thing(tm) to do. XGI, on the other hand, is still pulling the same crap SiS and Trident were so infamous for. Whee.
Both XGI and VIA do budget GFX cards with subpar performance for anything 3D. They're doing this as a marketing ploy.
Numerous posters have noted "why doesn't nVidia/ATI do this since we have to buy the HW anyway?"
Since the drivers communicate with the low-level hardware, they reveal partially the design of the hardware and I would assume there's some clever optimizations as well. They're not worried about Joe Sixpack having the source. They're worried about another card company looking at the code and thus making drivers/cards with better performance.
While open source is great for many things, I'm perfectly comfortable with not editing my video driver source code. Why do we really need it? It's not like this is a webserver, a browser, or anything I'd want to try to improve.
I'd much rather have these companies keep their drivers closed-sourced and use some of their profits to fund OSDN or the EFF.
-- Political fascism requires a Fuhrer.
Yes, Tim Roberts did a good job. But the free ProSavage (DDR) drivers are still limited. To date theres no working support vor XvMC for this card(s).
VIA S3 Savage can boot straight to TV. Use a VIA M10000 or similar Mini-ITX for your HPC and you'll be able to run Linux with open source drivers for everything, and boot to TV.
GCHQ Quantum Insert installed. If only our tongues were made of glass, how much more careful we would be when we speak
This is clearly too late: it just seems a move to get some attention for two families of GPU that did not receive much attention.
If VIA and XGI really believe in Open Source, they should give out the source and all needed documentation BEFORE the chipset hits the market, not after months or years.
This means having Linux/BSD/Else drivers ready when the chipset is finally available for the end user.
So can older ones - the 5200 in my Shuttle for example just detects the TV. Actually so does the Radeon 7000 I had in there briefly, too.
We had a rep from a taiwanese embedded mobo manufacturer at my workplace today trying to sell us boards (we make display equipment for bookmakers). He didn't know that VIA had brought out open-source drivers for their integrated graphics and was pleased to have another selling point. It would have helped a year ago when we were speccing the hardware, but there you go.
This is a welcome move in some respects, it certainly shows that VIA now considers the Linux user as a valuable customer base that must be supported.
However, there is already a thriving open source driver implementation for this platform: link providing code that was based on a version of VIA's code that they released to a limited number of open source developers a few years ago.
It is also worth noting that the "VIA Open Source" package still relies on a proprietary binary library to provide MPEG acceleration on their hardware. This library provides a completely non-standard API that applications must interface to to implement MPEG support. This is in contrast with the Unichrome project's solution, providing full source code for their MPEG implementation and implementing the standard and well established XvMC API in their driver.
The Unichrome project has also been responsible for implementing support for VIA MPEG acceleration in Xine, MPlayer , MythTV. Again this contrasts with VIA's solution to application support which has resulted in them producing forked VIA specific versions of Xine [VeXP] and MPlayer [VeMP] without involving the donor projects in the process or contributing back to them.
Along with many performance, feature and stability enhancements to the codebase the Unichrome project has also been responsible for reorganising and cleaning the codebase to a state that is now acceptable for inclusion in the base X.org source tree and resolving the security issues in the DRM code so that it can be included in the official Linux kernel.
It is, therefore, a shame that VIA decided to make this grandiose eye catching announcement, rather than just getting involved in the existing open source communities and simply helping and contributing to the Unichrome, Xine, MPlayer and MythTV projects. That might have been less eye catching or press release friendly, but it would certainly be a better way to win friends in the Linux community.
"Since the drivers communicate with the low-level hardware, they reveal partially the design of the hardware and I would assume there's some clever optimizations as well. They're not worried about Joe Sixpack having the source. They're worried about another card company looking at the code and thus making drivers/cards with better performance."
Your post is predicated on the assumption that:
1) Competition lacks the expertise to reverse-engineer either the software or hardware.
2) They're complete idiots when it comes to the principles, and mathmatics of computer graphics, and other aspects of hardware and software design.
3) That the computer graphics field is not as insular as people think it is.
4) Nvidia's actions as it relates to the nForce ethernet drivers.
In other words, we don't buy what your selling.
"While open source is great for many things, I'm perfectly comfortable with not editing my video driver source code. Why do we really need it? It's not like this is a webserver, a browser, or anything I'd want to try to improve."
DRM! Or do you need a worse excuse?
What difference does it make if drivers are open source or proprietary?
1st of all, Since nVidia (or ATI, or VIA) made the card, I'm pretty sure they know best how the hardware works and therefore they should know how to make the drivers.
2nd, I'd be scared of having some serious issues with OSS drivers. For the same reasons talked about above.
3rd, I don't care if the drivers are OSS, if it lets me use my device (not just vid cards, and drivers) and it's stable then what more do I need?
4th, isn't there a LSB on drivers? If there is, as long as they all meet the standards, why does X.org need the source code in order to use the video card "properly" sounds like X.org is the one with the issues.
5th, how many forks of the drivers would you be happy seeing? I mean there's like 8 Red Hat forks, so, and lord knows how many debian forks... would you like everybody to start making forks of the drivers? I'll re-code it so that left is really right, and up is really down, and call it the "Erehwon Edition"
DarkMantle I been bored, so I started a blog.
I don't know how to tell you this, but as predatory as it might sound, X.org's serfs are delighted with the potential for violent confrontation. I guess I should start by saying that the unalterable law of biology has a corollary that is generally overlooked. Specifically, I want to thank X.org for its principles. They give me an excellent opportunity to illustrate just how footling X.org can be. Although disorganized, eccentric big-labor bosses are relatively small in number compared to the general population, they are rapidly increasing in size and fervor. Let me carry my thoughts on this subject a bit further. X.org's compeers often reverse the normal process of interpretation. That is, they value the unsaid over the said, the obscure over the clear. Devious and temperamental, X.org's criticisms resemble a dilapidated shed. Kick in the door and the whole rotten structure will collapse, proving my claim that ever since X.org decided to go to great lengths to conceal its true aims and mislead the public, its consistent, unvarying line has been that its refrains prevent smallpox. If I am correct that X.org is talking out of its posterior, then people often get the impression that grotesque parasites and X.org's cringers are separate entities. Not so. When one catches cold, the other sneezes. As proof, note that X.org hates you -- yes, you, because you, like me, want to derail X.org's intrusive little schemes. Anyway, I hope I've made my point, which is that any effort to negotiate with X.org or appease it is akin to spitting into a hurricane to quiet its fury.
curse my lack of mod points.
"You know, Hobbes, some days even my lucky rocketship underpants don't help" -- Calvin
I think (hope) that's what that winking smiley ";)" was for...
unichrome project response
Looking on XGI's website: http://www.xgitech.com/about/about_hr_employee.asp
Company Age Distribution:
Below 20: 0%
20-29: 55%
30-39: 42%
40-49: 2%
50+: 0%
Would you call this ridiculous age discrimination?
So I guess I should shoot myself when I hit 40.
That I can now have acceleration on my laptop is just more sugary goodness.
After using NVidia on Linux, with binary driver installs (which works 95% of the time), and reinstalling after kernel update or new distribution, I gave up.
I switched to intel i845GV and i865GV chipsets. Completly open source drivers, included in distributions, 3D works out-of-the-box. Tuxracer, Enemy Territory, America's Army work great (I haven't tried UT). I love when stuff just works and I don't have to spend time to setup drivers.
I hope more producers will follow and we can finally stop paying for the CD with drivers and software included with hardware products (my motherboard has lots of drivers+norton antivirus, my dvd writer has Nero, I don't need them!)
Looks like my next graphics card will come from one of these two companies.
Get the floppy images, or the iso image for the latest stable release of FreeBSD (which is currently at 5.4 RC1) and do a small install including kernel sources and ports. Copy the example make.conf file in /etc, change the processor type in it to what you own, change the compilerflags to O2 adding -funroll-loops and -ffast-math (see google or yahoo!), follow the handbook on how to build and install world and kernel (pretty easy) and then start compiling KDE or Gnome (or whatever WM you want).
:-)
Go through the list of ports to see what you can add, if using Nvidia get the port of nvidia-driver.
First time I recompiled the whole thing on my laptop I was rather surprised and very pleased at how fast the thing is. It's only now that I found out how fast the thing really can be after using Windows on it for so long... Been tracking FreeBSD 5-STABLE since the release of 5.3 (used 4-STABLE for some time before) and I'm very pleased with the performance. (hoping to get a second laptop to follow progress on 6.x)
I can't force you to use it, but I think you should give it a try. The worst it'll cost you is a few days.
home
you can get updated drivers for your 3dfx card here . you can also get cool tips, tweaks and tools for various other things as well as Nvidia and ATI drivers(if you must). Game patches to run games such as Quake 3 and Doom 3 among many others can be found here . enjoy!
One thing that everyone appears to have overlooked is that gamers are fickle.
As such, if either Nvidia or ATI were to fall behind the competition from a performance standpoint by too large a margin, unless they continued to get good OEM shipments they would be doomed.
Hardware is only part of the problem in giving good performance, driver technology is attributable for much of the performance.
Neither Nvidia nor ATI want the competition to know what tricks they are up to inside their driver code.
Here's the quick summary:
*shakes fist at sky*
Free as in mason.
I've downloaded the source to the CLE266 video driver from the VIA Arena site under the Open Source section. It looks like they've removed the register names and bitfields for the 3D registers. Take a look at via_driver.c on line 5859 - that's no use to anyone who needs to work on the driver.
VIASETREG(0x43C, 0x00100000);
VIASETREG(0x440, 0x00333004);
VIASETREG(0x440, 0x10000002);
VIASETREG(0x440, 0x60000000);
VIASETREG(0x440, 0x61000000);
VIASETREG(0x440, 0x62000000);
VIASETREG(0x440, 0x63000000);
VIASETREG(0x440, 0x64000000);
How are we supposed to know what that does? It might as well be proprietary.
If the Nvidia driver were open source, I, as well as others, would be able to poke around in it and try to fix the bugs that Nvidia's engineers don't seem to care about right now. Then I can actually use the driver without it freezing up my machine. As it stands, the driver is closed source. So I am left to sit around and wait until Nvidia decides to fix my problems for me. In the meantime I get to use the unaccelerated slow Xfree driver that doesn't support all of my cards features. Do you understand the difference now?
The fairly cheap NVidia I have displays at 1024x768 scaled nicely to show on a TV which runs at a much lower real resolution. To show DVDs full screen it is rescaling twice, but looks very good.
The current FOSS ATI X drivers are based on information (and, OTTOMH, code) released by ATI and good for cards up to the 9200 series. Higher-powered cards will run also, but the more advanced hardware features aren't used.
NVidia didn't even release the source to a commodity item like their nForce LAN chipset, so we had to clean-room our own for that one.
The Volari cards look good. I'm pleased that the hard-working lab-rats there have finally managed to convince management to Open their 3D drivers too (they Opened their 2D stuff more than a year ago). Now all I need to do is fine someone in Western Australia who sells the XGI cards other than as a novelty item.
The VIA S3 cards suck. S3 cards have always sucked, from their horrid little every-one-different pre-Virge series on down. The CyberBlades sucked less, but were still not in the same league as their competitors, not even on par with Intel's basic integrated chipset. At least now we might be better equipped to work around some of the suckiness.
Got time? Spend some of it coding or testing
Also, to clarify on the S3 cards, the original Tridents didn't suck much more than their peers, but a lot of said peers went through revolutions and quantum improvements (and other got killed off), but the Tridents just missed out.
Got time? Spend some of it coding or testing
If they keep the driver under wraps you have to get another video card for the next release of windows without worrying about hackers modding the driver so they have to compete with their own and other legacy products.
Dear VIA,
Thanks a LOT. Thank you, thank you, THANK YOU for releasing open source drivers for the CLE266 chipset nearly two years after I first started purchasing these boards for use in my business.
Thanks for making me struggle for those years trying to install various proprietary, binary drivers on my "unsupported" Gentoo Linux workstations. Thanks for suggesting that I just use the abysmally slow VESA drivers for over a year on your forum. And thanks for letting me spend countless hours downloading the lastest Xorg tree from CVS so I could install reverse-engineered 2D-only drivers that finally made my systems functional, even if the more advanced features of your hardware have never, ever worked right.
I've also spent a lot of time investigating alternatives to your product, and I can say that at long last I've got a functioning, low-power, compact, ULTRA-QUIET Unix workstation, and all the hardware actually works. Yes, my new Mac Mini is a revelation, and I am about to deploy these new boxes throughout my shop.
Well, VIA, thanks for all the memories! My old Mini-ITX boards will soon start showing up on eBay, just in time for some new chump to try and install Linux on them using your new open source drivers. Maybe in another 2 years, when nobody really cares anymore, you'll release drivers for the accelerated MPEG hardware!
The chips and the linux boxes are out there in numbers.
Those ProSavage cards are commonly found in, inexpensive datacenter specials.
The OS/Chip ratio is good, and the numbers are pretty high in general, compared to 5 years ago when this kind of story would have been real news.
Still, I doubt that any of those machines would be used for their open source display capabilities.
Still good.
I didn't see any evidence that the 3D worked at all, especially with GLgears getting such terrible frame rates.
ATI's driver was the only one that allowed 3D accel, and it was the shits for everything.
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.