S3 Graphics Fails At Delivering Linux Driver

Ashmash writes "Phoronix is running a story about S3 Graphics failing to provide Linux support for their Chrome 500 products even though they have announced in press releases going back months that there is Linux support. S3 Graphics has gone as far as advertising OpenGL 3.0 support for Linux and one of their representatives had promised a driver by last December. This situation has been going on for months, but there is no Linux driver at all for the Chrome 500 series."

Comments

S3 is still around?

[Gojira Shipi-Taro]

This is news to me...

Pretty bad form to promise drivers and not come up with them. I wonder though, if their products are any good at all? Last S3 stuff (Virge, I think) that I saw was easily crushed by Nvidia and ATI.

Re:S3 is still around?

[drinkypoo]

I wonder though, if their products are any good at all?

The Original Virge considerably predates the existence of nVidia and when it came out there was nothing even close. It didn't really speed anything up, but it gave a substantial boost to visual effects. It actually tended to come at a frame rate penalty vs. software renderers except on the fastest machines.

Unfortunately S3 never really went anywhere after that, tried valiantly to go out of business several times, and mostly produces shitty integrated graphics. I just did a Windows XP install on a system based around an ASUS motherboard with a VIA chipset and S3 integrated graphics. They seem to work, that's about my only experience with them these days. But certainly S3 has no graphics solutions which will impress anyone. They are solidly at the bottom of the budget bin.

Re:S3 is still around?

[Vectronic]

I don't think they are really trying to participate in the gaming/high end of graphics, I think they largely still focus on onboard/handheld/as little as needed to work graphics.

Which, they are still pretty damn good at (usually), although you may know them better as VIA.

Re:S3 is still around?

[Roman Mamedov]

I'd consider S3 Savage and development of S3TC (licensed by Microsoft and others, renamed DXTC, still lives to this day in both ATI and nVidia cards) to be their moment of fame, and not Virge, which was also known as the first "graphics decelerator". :)

Re:S3 is still around?

[Anonymous Coward]

Personally, S3's biggest recognition in my life was playing Heavy Gear 2 at a LAN. Some putz was putting mortars on us from way the Hell at the far end of the map with impunity. After allegations of cheating and threats of beating, we all found out his Savage4 didn't support prerry much any way of rendering distance fog. He had an infinite viewline.

After further allegations of cheating and rigging, we found out that not only did he not intentionally arrange a corner case to get this screwup, there were no drivers from S3 that could fix it, and S3 had stopped making drivers for his card.

Re:S3 is still around?

[drinkypoo]

Nobody cares about S3 Savage3D. Even people who worked there didn't care (I used to know some.) S3TC didn't save S3 from becoming an also-ran. Texture compression was inevitable. Again, if you had fast hardware the Virge wouldn't slow you down... that much :) And if you tried to get the same results as the Virge (mostly lighting effects) in a software renderer, you'd see your frame rates drop one hell of a lot more than they did with a Virge. I would argue that the TNT is the first consumer graphics accelerator worth a crap, but the Virge did have its uses.

Re:S3 is still around?

[Daengbo]

The integrated low-end graphics are OK, unless you're using Linux and the driver leaves horizontal lines across your screen, leaving you with no option but the vesa driver (you can guess who's in this situation with a work computer).

TFA's report seems to show a big gap between what marketing wants to happen and what management is really doing. They're at least six months behind on the drivers. That's too bad.

Re:S3 is still around?

[__aamnbm3774]

i think that is the best story i've ever heard on slashdot.

Re:S3 is still around?

[0xygen]

I was cursed with a Savage4 after having a Voodoo 3 for ages... it quickly became known in our household as the Savage Whore, for the sheer number of times it got fucked one way or another.

The drivers were absymal, the hardware regularly locked up and it got replaced real fast.

Re:S3 is still around?

[Anonymous Coward]

The Original Virge considerably predates the existence of nVidia and when it came out there was nothing even close.

the original virge came out in 1995. nvidia was founded in 1993 and put out the nv1 in 1995.

I don't know how you can claim there was nothing close to the original virge, when software rendering would beat it.

there were a host of companies vying to put out 3d cards within the same window of time, including 3dfx (founded 1996), matrox (mystique in 1996), ATI (Rage2 was released in 1996, putting the Rage out in the 1995 time frame), Rendition (V1000 out in 1996), etc..

to suggest that the S3 virge "considerably predates" any of these is misleading, if not false.

Re:S3 is still around?

[jonaskoelker]

It actually tended to come at a frame rate penalty vs. software renderers except on the fastest machines.

Could you please explain this?

It sounds like you're saying that the hardware is slower than the software, except when the software is really really fast.

"This is Chewbacca, [wookie, endor] [...] this does not. make. sense!"

Re:S3 is still around?

[hairyfeet]

The S3s aren't SUPPOSED to impress anyone...that simply isn't there purpose. I don't know why folks always bring up Nvidia when talking about S3 because they really aren't the same market at all. It is like bring up the Core 2 Quad when talking about a Geode CPU. The S3 is made to be a LOW COST integrated graphics solution. I believe the latest Chrome 5 bumps it up just enough to run Vista with Aero, but gaming isn't something ever meant to do on these chips.

I can say that I have built several office machine running Win2K/XP on S3 and they work quite nicely for that task. They don't generate a lot of heat and thus you can build a nice quiet office machine with it. But it really isn't the same market as ATI/Nvidia. It is the same market as the Intel 945, that is a low cost budget GPU for rendering desktop graphics. Although I believe the last two generations of S3 support offloading some video as well. But for what they do they really aren't bad chips. But I have to say that it is bad form IMHO to promise drivers and not deliver. If I was a Linux user I probably would avoid the company in the future.

Re:S3 is still around?

[argiedot]

A similar, though less interesting, thing involving Counterstrike happened to me. On the old computers I had, it turned out that at the beginning of the round, every player would be displayed (normally, walking, crouching, shooting) but directly above their actual position. Sometimes this would go on throughout the whole game. It was pretty funny, I always knew where they were going.

Re:S3 is still around?

[FreonTrip]

What he's trying to say is that running in the software mode's default resolution of 320x200 was generally faster than trying to use the S3 Virge at 640x480, in spite of the performance-offloading and graphics-enhancing characteristics of the graphics chip. Further, a PC with an S3 Virge and a very fast CPU could provide a performance boost that would finally bump it consistently ahead of the software-only solution, but only just.

The Virge really could have been acceptable if its limitations had been taken into account - essentially writing a low-level rasterizer for the chip that equaled software quality would have provided a noticeable performance boost - but when the chip was also expected to handle z-buffering, bilinear filtering, perspective correction, and mipmapping in addition to triangle setup and polygon painting, it was simply overwhelmed. It's the fate that befell many, many first generation 3D accelerators - the Virge was just at a weak position even within the company of its benighted brethren.

Re:S3 is still around?

[Kazymyr]

The Original Virge considerably predates the existence of nVidia and when it came out there was nothing even close. It didn't really speed anything up, but it gave a substantial boost to visual effects. It actually tended to come at a frame rate penalty vs. software renderers except on the fastest machines.

Yeah - S3 Virge/DX, the video decelerator. I still have one in a box somewhere. In action games you were better off using software rendering, as you'd get more FPS.

Re:S3 is still around?

[Eil]

I would argue that the TNT is the first consumer graphics accelerator worth a crap, but the Virge did have its uses.

I guess it depends on what you define as "worth a crap," but it was the 3Dfx Voodoo chipset and GLQuake that *really* launched the PC as a serious gaming platform (and the six-month upgrade cycle that goes with it). Back in the 90MHz Pentium days, rendering a 30fps 3D scene at 640x480 with consumer-level hardware was no small accomplishment.

Re:S3 is still around?

[dotancohen]

Some putz was putting mortars on us from way the Hell at the far end of the map with impunity.

Had that happen to me in Lebanon for a few days. Now I know how they did it!

Re:S3 is still around?

[drinkypoo]

I guess it depends on what you define as "worth a crap," but it was the 3Dfx Voodoo chipset and GLQuake that *really* launched the PC as a serious gaming platform

This is a completely fair assessment. It was also something of a nightmare. I was completely blown away with it, but it was a serious annoyance, so serious that I actually bought a PowerVR. I think I still have it, although I don't think I've been static safe since... Once the TNT came out I never looked back. I escaped ATI's 3d stuff entirely (but had had plenty of problems just with Mach32 and Mach64 inconsistencies already) until I got a laptop with Rage Pro, of which it can be said that it is not pure trash. I had a Permedia2 card next which came at a steep price premium and had slightly less performance than a Voodoo 2 card but had about 56% less hassle - and which had real working OpenGL. But that was really a low-end pro card. I had a Riva128 for a moment, subjected it to defenestration when the TNT came out, and have never looked back - nVidia FTW!

nVidia has pulled some serious boners over the years but in general they have delivered the most workable 3d solution, cross-platform and all. I had just discovered what it was like to have money when this stuff was coming out, so I had most of it. I never spent much money on a CPU, so I was pretty familiar with which video cards were more and which were less CPU limited, which was actually a major feature of the 3dfx stuff at the time. But I still curse their name eternally for inventing GLIDE instead of just starting with MiniGL, which would have been a much kinder thing to do to the gaming industry and probably would have resulted in a world without Direct3D.

We can dream, can't we?

Re:S3 is still around?

[i.of.the.storm]

My experience with S3 has been that their drivers absolutely suck, and while they might have some decent IGPs (not the one I had, certainly) it's probably better to get something with Intel, ATI, or nVidia integrated just so that the drivers aren't utter horseshit. I actually knew someone who worked at S3, and he got me a free graphics card which was somewhere between the Radeon 9600 and 9800 in terms of theoretical output, but it crashed so much that I went back to my lowly 9200, which was still an order of magnitude better than the S3 ProSavageDDR IGP that was originally on the machine.

Re:S3 is still around?

[Anonymous Coward]

Actually the Matrox Millennium, which was released first, crushed the S3 Virge in performance.

Re:S3 is still around?

[francium de neobie]

I've had a Savage4 before and it absolutely sucked - tons of z-buffer artifacts, buggy drivers and bad performance even if the driver works.

Re:S3 is still around?

[hairyfeet]

That is why I don't usually use the drivers for the S3s. The drivers in WinXP usually perform well enough for office work that there simply isn't a need to bog it down with pointless code that will never be used and add instability anyway. I feel the same way about budget AC97s and the SiS network cards. Often the Windows drivers simply work better and are more stable than those put out by the actual manufacturers. And don't EVER use the SiS IDE drivers, ever! They will slow down the IDE drive and make it a hell of a lot more unstable than the default Windows drivers.

I have found that for a lot of these budget parts the Windows drivers are more than "good enough" for what most folks are going to be doing with it and it just isn't worth risking stability for a little bit extra performance. And in the case of the SiS you may actually get less performance than by sticking with Windows defaults.

Re:S3 is still around?

[shutdown -p now]

The S3s aren't SUPPOSED to impress anyone...that simply isn't there purpose. I don't know why folks always bring up Nvidia when talking about S3 because they really aren't the same market at all. It is like bring up the Core 2 Quad when talking about a Geode CPU. The S3 is made to be a LOW COST integrated graphics solution.

The thing is, both NVIDIA and ATI also have low cost integrated graphics solution offerings, and they are seen much more often than S3. Last time I shopped around for a prebuilt PC (which was like a month ago), I didn't see a single one with S3, cheap or not.

Re:S3 is still around?

[i.of.the.storm]

Yeah, I usually do that these days too, for everything except for my graphics card. It's actually surprising how many devices work out of the box with Vista, and almost everything else I can pull the driver off of Windows Update. But back then that was the only computer in the house and I wanted to play my games, so I had no choice but to try and mess with the drivers.

I think the best thing to do though is just to go with medium-high quality parts from reputable brands, and you don't have to worry about stability as much. I usually buy stuff that's just a little more expensive than the cheapest item on newegg, and it seems to serve me well. There isn't much reason to use SiS or VIA chipsets these days, which is bad in terms of competition but those are the facts. I remember ther was a time not too long ago when VIA boards were still competitive, but these days it seems like between Intel, nVidia, and AMD chipsets, you don't really need to look elsewhere.

Re:S3 is still around?

[Firehawke]

I can summarize easily enough, having had one of these cards in my possession for a short period of time.

Running any sufficiently modern (for that time) game at 640x480 was faster in software rendering than it was with the hardware rendering and often looked better too.

Re:S3 is still around?

[muchenik]

Problem is that for $70 you can get this card and for $30 you can get the ATI HD3450 or for $50 you can get the nvidia 8400GS. All cards run within 10% of wattage and have similiar performance and all are multimedia cards. The ATI HD4350 seems to be the answer for sub $50 and the NVIDIA 9400GT seems to be the answer for the sub $100. So really if they would have been able to put out at lower wattage we may have been able to see numbers on a PCIE slot. Now I would like to see them take what they have learned here and apply it back to their integrated chips.

WTF.

[MukiMuki]

If a driver isn't out on day one, there's no way in hell this should be in a press release. I can only hope that it doesn't make it to any of the boxes.

Bullshit like that shouldn't be legal.

Re:WTF.

[drinkypoo]

Bullshit like that shouldn't be legal.

Mistakes are made. On the other hand, you and all the other people who bought one expecting Linux support (you are one of those people, right?) can get together and get a class action suit going against them.

Re:WTF.

[Daengbo]

Mistakes are made, but we're still waiting for the last chip's announced driver, and they're announcing the new chip (also w/ non-existent driver). That's a pretty huge mistake.

Re:WTF.

[pdabbadabba]

Bullshit like this *isn't* legal. See, for example, CA's Consumer Legal Remedies Act: http://www.harp.org/clra.htm

For example, it is illegal under the CLRA to:
"(5) Represent that goods or services have sponsorship, approval, characteristics, ingredients, uses, benefits, or quantities which they do not have or that a person has a sponsorship, approval, status, affiliation, or connection which he or she does not have."

(IANAL)

S3 has always been a synonym for "avoid"

[Alioth]

For Linux users, even back to the early 1990s, S3 has been a synonym for "don't buy this graphics card". Even back then, they didn't release specs for their graphics cards, and they didn't even support VESA modes for graphics mode so their cards couldn't be used at all for X.

At least the other two closed graphics cards makers do supply drivers for Linux.

Re:S3 has always been a synonym for "avoid"

[morgan_greywolf]

Yep. Wayyyy back when, when I first tried Slackware and couldn't get X to work with my S3 graphics card, my posts were answered with something along the lines of "Get a Riva TNT or an ATI card."

Re:S3 has always been a synonym for "avoid"

[roystgnr]

For Linux users, even back to the early 1990s, S3 has been a synonym for "don't buy this graphics card".

"For Linux users"? I've never heard of any Windows user intentionally buying one, either... the ones who've heard of S3 know them as the company which once marketed a "3D decelerator", a card so slow that a new computer would be better off with software rendering.

Re:S3 has always been a synonym for "avoid"

[Antique Geekmeister]

Interesting. It sounds like Promise's Linux support: they used to publish customized versions of out-of-date versions of Linux patches that re-arranged your drive numbering without telling anyone and coulldn't be applied against any contemporary kernel source tree.

What NVidia publishes is a fascinating attempt to endrun around the GPL licenses. They publish a binary blob kernel driver, which 'taints' your kernel and legally prevents you from being able to publish it as part of your distribution. And for the features to work, they also replace the OpenGL libraries with closed source versions, which do not integrate well with _anyone's_ package management system, because they refuse to publish any RPM, .dep, or other package managed tool for it, nor does their license allow repackaging it and automating installation. If you want the drivers, you're compelled by their license to _manuall_ have each installer sign the agreement.

This is extremely painful in many environments, and leads to some extremely poorly done repackaging of their drivers and libraries for automated installation, and their exclusion form Linux distributions.

Re:S3 has always been a synonym for "avoid"

[cdrudge]

Interesting. It sounds like Promise's Linux support: they used to publish customized versions of out-of-date versions of Linux patches that re-arranged your drive numbering without telling anyone and coulldn't be applied against any contemporary kernel source tree.

<tongue-in-cheek>
That's the advantage of using Linux! Anyone can take that source and update it. You have an entire community to help make sure that it works and to maintain it. Don't blame the manufacturer for lack of community support of their device.
</tongue-in-cheek>

Re:S3 has always been a synonym for "avoid"

[canavan]

My experience from 1996 is quite the oppiosite. I bought a S3 964 based card after those were on the market for more than a year, and I had to find that XF86 was running in false-color so to say. S3 sent me the printed programming for free by international airmail essentially no questions asked.

It turned out that the problem was the way the external IBM RAMDAC was wired to the S3 chip - easily fixed with a 2 or so lines patch, Back in those days manufacturers of graphics cards dodn't just implement a reference design, but had lots of choices which components to connect in which way, which didn't really make things easier for open source developers without access to every hardware variant out there.That essentially ended with the S3 Trio/Virge and the ATI and NVida 3d cards.

Re:S3 has always been a synonym for "avoid"

[John Hasler]

Except that the patches existed to make the kernel compatible with their closed-source code. Without the source for that code there was no way to know why the patches did what they did and so no way to safely "update" it.

Re:S3 has always been a synonym for "avoid"

[Ant P.]

I accidentally bought a Savage4 once, because the only other thing the shop stocked were low-profile-slot GeForce 440MXs. It had problems running 2D, let alone 3D apps.

Re:S3 has always been a synonym for "avoid"

[Antique Geekmeister]

Oh, my. Tongue out of cheek, the public Linux drivers were in fact already in the main kernel source tree, and were far better than Promise's 'Linux support'. Promise is famous for many, many other manufacturing and design problems, so it's not a big surprise. It looks like S3 has similar historic quality problems.

Re:S3 has always been a synonym for "avoid"

[drinkypoo]

Yeah, you fucked that decision up. Next time get the low-profile card, bend the frame straight, and secure it with tape. GF440MX was a peach, sure it was slow, but so solid.

Re:S3 has always been a synonym for "avoid"

[PitaBred]

Actually, NVIDIA specifically allows Linux/BSD repackaging with relative impunity:

2.1.2 Linux/FreeBSD Exception. Notwithstanding the foregoing terms of Section 2.1.1, SOFTWARE designed exclusively for use on the Linux or FreeBSD operating systems, or other operating systems derived from the source code to these operating systems, may be copied and redistributed, provided that the binary files thereof are not modified in any way (except for unzipping of compressed files).

I still don't like the binary blob approach, but they aren't actively blocking any redistribution. They're encouraging it. It's just that distros like Debian don't use it because they have a (valid, IMHO) philosophical argument against binary blobs.

Re:S3 has always been a synonym for "avoid"

[Jeffrey Baker]

This doesn't seem right. The early 90's S3 card were some of the easiest to get working in XFree86. The Trio32, Trio64, 968, etc were all extremely common in those days. They were marketed under popular brands like Diamond, Number Nine, Hercules, and STB.

Re:S3 has always been a synonym for "avoid"

[Alex Belits]

S3 sent me the printed programming for free by international airmail essentially no questions asked.

Was this sentence supposed to contain the word "manual" somewhere?

Re:S3 has always been a synonym for "avoid"

[cibyr]

He accidentally the whole printed programming.

Re:S3 has always been a synonym for "avoid"

[damium]

There was a patch (not by S3) for the XFree86 driver that disabled VESA mode probing. It would make S3 chipsets "work" with VESA drivers if you got your mode lines just right in the config file. Even then there were some other bugs, you had to start the kernel in framebuffer mode and you could not go back to a console once X was started.

The stated reason from S3 for not supporting VESA probing was that it was used in the past to obtain firmware from the chipset and that Windows (9x) did not probe for VESA modes (even when using VESA drivers).

I like to think that linux support from manufactures has come a long way since then but it looks like some still don't care.

Re:S3 has always been a synonym for "avoid"

[Z80a]

4MX was quite insane in T&L speed by the price of it, not many cards of this price range can run that 1 million of triangles 3D mark test at 20-30 fps

Maybe it's time to help out

[morgan_greywolf]

It may be time to help out The Unichrome Project, which produces a driver that works with the older Unichrome and Chrome9 chipsets.

VIA doesn't have much of a history of helping the open source community with specs or source for its S3 graphics cards.

Re:Maybe it's time to help out

[Jeff DeMaagd]

I thought Via was more known for making crappy products that Linux users shouldn't be buying in the first place. Even if the drivers are good, the hardware was often marginal or bad.

Re:Maybe it's time to help out

[DJRumpy]

I'm a little confused as to why S3 isn't more open with the Linux community? They are obviously not top of the line and the technology they use isn't going to be snapped up by nVidia or ATI. I doubt that the big two are beating down their doors trying to plant industrial spys. Why won't they release specs? I know nothing about graphics hardware. Would releasing the specs expose them to corporate espionage of some sort?

Re:Maybe it's time to help out

[John Hasler]

> Why won't they release specs? I know nothing about graphics hardware. Would releasing
> the specs expose them to corporate espionage of some sort?

It might expose them to lawsuits and/or cancellation of licenses. Many of these outfits don't own all of the "IP" in their chips and/or drivers. They license it from other companies and in doing so agree to some truly amazing restrictions.

Re:Maybe it's time to help out

[DJRumpy]

So I take it the bigger shops like nVidia and ATI typically either have better lawyers or do more of their product in-house? I suppose that makes sense. Sucks all the same though.

Re:Maybe it's time to help out

[FishWithAHammer]

3) Patent encumbrance

Re:Maybe it's time to help out

[argiedot]

I hate Via with a vengeance, their goddamn Unichrome cards are just a goddamn pain. They used to provide drivers which would either not compile on your kernel, or induce a kernel panic on an older version. Fuck them.

Re:Maybe it's time to help out

[morgan_greywolf]

*shrug* My mainboard is a VIA K8M800 chipset motherboard. What can I say? It was cheap. Runs good though. I don't use their crappy video chipsets, though, I usually stick with Nvidia for that. I encountered the Unichrome driver trying to get my wife's onboard Chrome 9 graphics adapter (K8M890 chipset) to work and after playing with it for a few hours decided to just say 'screw it' and got her an Nvidia Geforce 6200 LE instead.

I will say that for some reason the onboard VIA Rhine III Ethernet controller was junk on that same board. It would randomly drop packets, etc. The add-on card I put in there happens to be a VIA Rhine II (same as on my board), and it works just fine.

Anyway, YMMV.

Re:Maybe it's time to help out

[morgan_greywolf]

The Unichrome Project is completely unrelated to VIA. Just a guy with some Unichrome cards and the community working together to write open source drivers. Don't blame the poor guy, it isn't his fault.

Re:Maybe it's time to help out

[argiedot]

I actually find the work done at The Unichrome Project and at the OpenChrome project incredibly helpful and can't recommend the work done there enough. 'Unichrome' is the name of a series of IGPs released by Via that they put on their motherboards. At one time (I don't know if it's still there) Via had driver source code on their website that would have the effects I mentioned earlier. Funny, no? Better drivers from reverse-engineers than from the company. Tells you that the company is crap.

Re:Maybe it's time to help out

[MrMacman2u]

4.) Laziness and $$. Support is hard work and expensive.

linux graphics cards

[Janek Kozicki]

So far, currently graphic card is the only serious issue on linux, for me. I'm happy that AMD released open source ATI drivers, but still they are crap. When university asks me what hardware to buy new students for working with yade I tell them - whatever, but make sure that it has nvidia card and shh.. you could consider AMD too ;)

Sometimes a clueless grad student comes and is wondering why yade works like crap, and I see instantly - it's the graphics card. Good for us, that all serious computations are done remotely on a cluster without graphics at all. OpenGL is used only for setting-up the calculations etc..

Re:linux graphics cards

[Narishma]

AMD didn't release open source drivers. They only released the specs of some of their cards.

Re:linux graphics cards

[i.of.the.storm]

I think they're also working with Novell to make the RadeonHD drivers though, or something like that. They're still putting out the closed source drivers of course, but I think they're also helping with the open source ones.

Re:linux graphics cards

[sricetx]

I considered AMD graphics. Specifically integrated 2100 graphics on a AMD740G based motherboard. I thought it would be a good replacement for the Nvidia FX5200 AGP card I was using in my previous build. I soon found out that all available drivers in Kubuntu 8.10 suck for this chip. The binary drivers from AMD are probably the most full featured of the bunch, but tended to crash when playing HD video, etc. Needless to say, I bought a cheap fanless Nvidia 7200GS PCI-E card for the new MB, and it works flawlessly.

S3?

[Anonymous Coward]

why are you even bothering buying S3?
When you could buy a card that has more support in the community. Thats like buying a winmodem and crying the company hasn't released a driver for it yet even though they said they would.
I learned my lesson about hardware and *nix, either buy what is supported or throw it away :P

Re:S3?

[LDoggg_]

For the little micro PCs that have this crap embedded instead of a PCIe or AGP slot.

Can we not just get fully open specifications?

[jamesmcm]

We really need fully Open specifications and drivers for graphics cards. ATI has released some specifications but have kept their drivers proprietary - greatly increasing the workload for the community.
Nvidia have only produced proprietary drivers and published no documentation.

We really need a company to publish all the specifications and produce GPL-compatible GNU/Linux drivers, that way the community could work on improving htem and their features could be fully utilized (in video playback, etc.). It's a win-win situation, the company has help producing the drivers and the community gets better drivers.

Re:Can we not just get fully open specifications?

[quanticle]

As I recall, ATI opened the source and specifications for their drivers on September 10, 2007. How do you figure that ATI is still a closed standard?

Re:Can we not just get fully open specifications?

[jamesmcm]

I wasn't aware they had open-sourced their drivers as well. ATI is definitely the way to go then.

Ah, it seems the older drivers are proprietary.
From wiki:"Linux users have the option of both the old proprietary (R200 and above) and new open source (R480 and below) drivers."
It's nice to see the newer ones are Open Source though.

Re:Can we not just get fully open specifications?

[Xabraxas]

We really need a company to publish all the specifications and produce GPL-compatible GNU/Linux drivers,

We already have that. It's a little company called Intel

Re:Can we not just get fully open specifications?

[jamesmcm]

Yeah, but they're not mass-producing medium-high end cards. They tend to just make integrated chips (do they make any separate cards?) for use on netbooks, etc.

Really we need Nvidia to do it, as it's the only way GNU/Linux will get better compatibility with the normal desktop. (ATI is already playing ball)

Re:Can we not just get fully open specifications?

[Haeleth]

Intel graphics chips are fine for pretty much any Linux purpose -- Compiz runs fine, KDE4 runs fine.

They can't compete with NVidia or ATI for playing the latest 3D games, but that's unimportant, because 99.999% of the people who care about the latest 3D games are either playing them on a console or in Windows.

Re:Can we not just get fully open specifications?

[Xabraxas]

Most people don't need high end graphics. Intel is more than enough for 99% of users. The other 1% is mostly gamers and most of them aren't gaming on Linux. It would be nice to have good open source drivers from both ATI and Nvidia but it isn't essential. There are more computers out there with Intel graphics than any other graphics manufacturer.

Re:Can we not just get fully open specifications?

[jamesmcm]

I agree, but most computers have ATI and Nvidia cards. Therefore it is important for hardware compatibility.

Re:Can we not just get fully open specifications?

[i.of.the.storm]

Mmm, unfortunately, not quite. They are helping to make the new RadeonHD open source drivers, but they are also still putting out the closed source drivers, and I think the open source ones are probably not as feature-complete as the closed-source ones. But it's still a good initiative from them.

Re:Can we not just get fully open specifications?

[jedidiah]

Intel has very respectable Linux drivers. ATI is at best a
distant third when compared to Nvidia and Intel.

Intel and Nvidia both have lowend embedded GPUs that are very
usuable for a wide range of tasks including intensive multi-
media. ATI just sucks.

Re:Can we not just get fully open specifications?

[Tacvek]

The latest 3d games? Try 4 year old games that were not even grpahically top of the line at the time of release. My younger cousin received a laptop for Christmas with the Intel GMA 950 chip. It was unable to play The Sims 2 without issue, which as you can guess is not the most demanding of 3d games. I was not surprised that some of the pixel shading effects were omitted, but it visibly had trouble keeping a framerate of 25. Not to mention the game crashes when using that chip. (This is probably a driver issue, with the driver corrupting game memory when trying to load textures, or something of the sort.)

Granted that that chip is known for being crap, and Intel offers much better ones.

(On the other hand, by ATI card is known to have periods of time on some games (Half-life 2 Ep 1) with a frame rate less than 5. In s few cases the frame-rate dropped below 1 fps! I suspect it was either slightly defective, or Windows was at fault here. )

Re:Can we not just get fully open specifications?

[Xabraxas]

I agree, but most computers have ATI and Nvidia cards.

Intel has nearly 50% of the marketshare, much bigger than either ATI or Nvidia.

Re:Can we not just get fully open specifications?

[ploxiln]

Yes, intel graphics (except for GMA500) and wireless chips are very well supported for linux in every sense.

Unfortunately, one cannot just buy any of these chips in a form which can be added to an existing laptop or desktop, they only come built into a motherboard or laptop.

Let's raise this barn!

[malevolentjelly]

I think this simply calls for a really long, over-promising and under-delivering open source driver project in the tradition of Nouveau or anything in DRI produced without commercial support. I look forward to a series of unstable and unusable "releases" which may someday, years from now, result in a shoddy but roughly functional driver.

Maybe some *prominent* linux developers should take some time out of their respective minor IT and sysadmin jobs to create and fully support their very first OpenGL 3.0 driver for this moderately unpopular architecture.

It'll be like a Little Rascals movie, but with more facial hair.

Re:Let's raise this barn!

[thsths]

> I think this simply calls for a really long, over-promising and under-delivering open source driver project in the tradition of Nouveau or anything in DRI produced without commercial support.

As much as I am skeptical of them, Intel seems to be the only company interested in open source drivers. ATI may be making moves in that direction, too, but I am still waiting for results.

As for VIA/S3: at least the specs for some of the chips are out now. Unfortunately, the existing drivers are still some of the worst code I have ever seen - and most things work more by accident than by design, if at all.

Re:Let's raise this barn!

[Kjella]

As much as I am skeptical of them, Intel seems to be the only company interested in open source drivers. ATI may be making moves in that direction, too, but I am still waiting for results.

They've actually made very big progress on the release of specifcations, so it's more on the open source side now. The last release of documentation was the R6xx/7xx 3D documentation that arrvied just a few weeks ago, but there's been some demo code released before that. Previously there's been releases of the ISA and various other bits, so people have been working on the drivers for a while. As far as specs is concerned having power management (basic suspend/resume should already be covered, but not during operation for mobile chips etc.) and UVD (hardware decoding of H.264/VC-1). However, all the specs required to write kick-ass 3D acceleration should now be there.

Unfortunately, one of the things messing things up is that the X graphics driver model is undergoing big changes with GEM (kernel graphics memory manager), KMS (kernel mode setting), DRI2 (new direct rendering interface), Gallium3D (low-level API for shaders) and to support hardware acceleration while using a compositing window manager. It's pretty much all necessary to bring the Linux desktop up to speed but it does throw a lot of balls up in the air at once. GEM was introduced in kernel 2.6.28 and is quite new, Gallium3D was just merged to Mesa master and overall most of this only lives in special compile flags and git branches. There's a lot of promise for 2010 but 2009 will be rough waters.

So what you're saying is...

[Anonymous Coward]

"Chrome still doesn't support Linux"

No surprise there.

Terrible news...

[dimethylxanthine]

This just ruined my Sunday afternoon.

Turn that frown upside down

[binarylarry]

This may be a failure for S3, but I think I speak for the entire Linux community when I say:

One less S3 chipset on Linux is a win for most Linux users.

Hopefully there won't be a successor.

And the problem is?

[cjjjer]

So yeah they probably should not have had the press release, but what some of you people don't realize is that the economy is tanking for alot of tech companies. Lets see, support product(s) that probably have less than lets say 10% of a market share, cater to group (linux users) within that product market lets say 1% at most. As you can see they probably don't really care to piss off 1% of users to help keep their company profitable.

Makes sense to me...

Re:And the problem is?

[enrevanche]

By doing this kind of thing, they screw people who would not have considered buying there product. Some of these people cannot afford to just get something else. This is false advertising and they should be held liable for it.

DDK emulator?

[pondermaster]

Maybe it's time for someone to write a DDK emulation layer? Should be able to yank out most of it from the ReactOS project. Not perfect, but at least the binary drivers from windoze could be use as a last resort.

Re:DDK emulator?

[LingNoi]

You mean the same ReactOS that had to stop and do a complete code cleanup after idiotically putting in tons of patented and reserve engineered code? No thanks.

Re:DDK emulator?

[pondermaster]

They did the cleanup years ago, so your point is moot.

Download,com

[ciderVisor]

Wow, the comments at the end of that article are priceless. Makes this place look like a debating society.

Re:Download,com

[DrSkwid]

I read the comments and now I'm stupider than when I started.

Deceptive advertising...

[gweihir]

... which I thought was illegal almost everywhere. Maybe the land of the class-action suit already has a solution in place to deal with this type of fraud?

Re:Deceptive advertising...

[John Hasler]

Have you filed a complaint yet?

Re:Deceptive advertising...

[gweihir]

And how would I do that, being European?

Why don't hardware manuf. just drop the specs?

[thesandbender]

I've never understood why video card manufacturers play there cards so close to the vest. The magic sauce is in the hardware... not the API that interfaces with it. Yes... the API gives some insight into what the hardware is doing... but not enough to reverse engineer the product.

Re:Why don't hardware manuf. just drop the specs?

[John Hasler]

> Yes... the API gives some insight into what the hardware is doing... but not enough to
> reverse engineer the product.

They may have licensed some of that magic sauce from other companies under terms that their lawyers interpret as barring them from releasing the API.

Re:Why don't hardware manuf. just drop the specs?

[hairyfeet]

Probably because of the giant minefield we here in the US have in regards to patents and copyrights. After all there are only so many ways to render a picture on a screen and thanks to the minefield anything you do is probably already in the portfolio of your competitors or worse a troll. So I really can't blame most of these companies for not wanting to risk a lawsuit for a small section of the market.

Of course this is a good example of why we need patent and copyright reform, because anyone making a product today faces a giant minefield of patents and copyrights simply to get their product to market. So the more things they can keep "secret" the less they have to walk through the minefield.

Re:Why don't hardware manuf. just drop the specs?

[LingNoi]

By doing it puts the barrier to entry into the graphics market extremely high.

Not only do you have to create a good card, you also need good software to go with it.

Who cares about S3 graphics

[Wolfier]

What's their top-of-the-line? Can it beat a mobile Geforce 7300?

Re:Who cares about S3 graphics

[Anonymous Coward]

yes i think it can... the S3 Chrome 530GT is about equivalent to a 8400 GS

really they aren't sold as gaming cards they are for HTPCs pretty much video accelerators vs 3D accelerators

Re:Who cares about S3 graphics

[BikeHelmet]

Yes, they've had quite good video acceleration for a while. They've had MPEG2 acceleration for years.

It's too bad that without drivers, acceleration does you no good. :P I mean sure it'd be nice to have MythTV accelerated... but it'd be nicer to have it run. ;)

Where they went wrong...

[w0mprat]

They should have developed with Linux as their primary platform with fully open source drivers with GPGPU stream processing and everything, because that is a untapped niche. Because S3 sure as hell can't compete with Nvidia/AMD on outright performance even when you factor in cost.

Judging by the framerate...

[dimethylxanthine]

...of the S3 banner from that link, I won't be getting an S3 any time soon...

Wow.

[M1rth]

You actually found enough people gullible enough to buy Heavy Gear 2 so that you could make a LAN game out of it?

That's a sad waste of money.

XvMC? openchrome , ATI, Open Graphics Project

[Something Witty Here]

> As I recall, ATI opened the source and specifications for
> their drivers on September 10, 2007 [linux.com]. How do
> you figure that ATI is still a closed standard?

ATI has not released docs for UVD or UVD2.
I haven't seen any docs for RAGE* or FIRE* chip families.
There is no FLOSS support for XvMC on any ATI chip.
I can't even get their stupid chip to do sync-on-green.

And of course nvidia is completely useless.

Back to VIA/S3...

So the openchrome(4) driver doesn't support this Chrome 500
thingie?

Does anyone make a PCI or PCIe card that the openchrome
driver supports?

The Open Graphics Project is working on designing a
completely documented video chip. The project could
benefit from additional engineering talent, and of
course, money. ASIC masks are very expensive.

Re:XvMC? openchrome , ATI, Open Graphics Project

[stinerman]

Wow! Someone else who could use XvMC. My aging Athlon XP 2400+ has trouble keeping up with 1080i MPEG-2 broadcast streams.

AFAIK, XvMC will never be implemented on any of the Radeons because they're going to move to the Gallium3D framework. Or they might go to VA API. Or something else entirely.

That's the problem: there are too many "solutions" to do hardware offloading for MPEG-2/4. There is no consensus about which to use, so no one is going to implement any of them.

VIA/S3 docs

[Something Witty Here]

For those interested, some docs:
ftp://ftp.vtbridge.org/Docs/

Re:Obama fails at delivering stimulus package.

[silent_artichoke]

I think it just a touch more likely that they just feel that this has nothing to do with S3 chips and Linux drivers.

Re:why are there so many computer illiterates here

[westlake]

chrome 500 series is basically a chrome 400 with a new display interface (hdmi/dp).
so all you are missing if you use the 400 series driver is DRM... who in their right mind would call that a problem..

anyone who can see that HDTV display and the single-cable HDMI solution for digital audio and video is the future of the PC in the home market.