VIA Will Not Provide An OSS Chrome 9 3D Driver

[phoronix]

Phoronix: VIA Will Not Provide An OSS Chrome 9 3D Driver

This morning VIA's Bruce Chang had submitted the DRM code for the Chrome 9 IGP with a request that it be pushed into the mainline Linux kernel. The DRM alone isn't usable to an end-user without a Mesa driver or something else to take advantage of this kernel component. VIA previously expressed interest in a Gallium3D-based Chrome 9 driver, but now today we find out they have no intentions on creating an open-source Chrome 9 3D driver. Instead, they just want this DRM into the mainline Linux kernel so that it can be used by their binary blob and to hopefully have some open-source developers come along and create a free software driver from their incomplete documentation.

http://www.phoronix.com/vr.php?view=14051

[Adarion]

Oh fuck you, VIA. A binary what? That strange thing of bytes that will only work (or better expressed: not work) on few selected age old distributions? You want to give us that?
Nah, forget it. I will not buy any piece of VIA's shoddy hardware, make sure not even to get soundchips or networkchips, unless this changes. It's ok to just give specs as Greg KH said and the community will care for a free driver but it needs at least these specs!
And telling us that one of the once largest chipset maufracturers isn't able to give at least a starting kick to it? Not havin the resources? Who in the fricken Windoze world buys VIA's shit? I doubt a single person would use a VIA box for gaming. They should rather make themselves a safe place in the Linux corner cause they won't stand long in the rest of the world.

[[Knuckles]]

Oh fuck you, VIA. A binary what? That strange thing of bytes that will only work (or better expressed: not work) on few selected age old distributions? You want to give us that?
Nah, forget it. I will not buy any piece of VIA's shoddy hardware, make sure not even to get soundchips or networkchips, unless this changes. It's ok to just give specs as Greg KH said and the community will care for a free driver but it needs at least these specs!
And telling us that one of the once largest chipset maufracturers isn't able to give at least a starting kick to it? Not havin the resources? Who in the fricken Windoze world buys VIA's shit? I doubt a single person would use a VIA box for gaming. They should rather make themselves a safe place in the Linux corner cause they won't stand long in the rest of the world.

Fully agreed. No one I know has a good opinion of VIA's hardware, and with the intel atom there is basically no reason in going for a VIA platform.

[mendieta]

;82035']Fully agreed. No one I know has a good opinion of VIA's hardware, and with the intel atom there is basically no reason in going for a VIA platform.

Yep, I am always voting with my wallet. My next purchase will be an ATI discrete card to celebrate the 3D Open Source drivers whenever that happens.

BTW, they could outsource to Tungsten Graphics to do this, no? It should be so much easier now with Gallium!

[Nille]

If via release the rest of the documentation i have no problem with an closed source mesa and DDX and an open source kernel module.

[Ex-Cyber]

Who in the fricken Windoze world buys VIA's shit? I doubt a single person would use a VIA box for gaming. They should rather make themselves a safe place in the Linux corner cause they won't stand long in the rest of the world.VIA hasn't been seriously pursuing the gamer/enthusiast market for years; they've been setting the standards for small, low-power PC systems. Intel and Nvidia are just now getting around to producing stuff that can compete with VIA in this space.

[Louise]

Have any one counted how many times the article repeats itself? :)

[mendieta]

Have any one counted how many times the article repeats itself? :)

Yep :D

I was missing a final paragraph:

I REPEAT. Via is not ...

LOL!

[Louise]

I didn't quite understand, was it the complete programming documentation VIA had released? :D

Well, at least they will release an Open Source 3D driver, once the DRM is merged to kernel :D

[Adarion]

VIA hasn't been seriously pursuing the gamer/enthusiast market for years; they've been setting the standards for small, low-power PC systems. Intel and Nvidia are just now getting around to producing stuff that can compete with VIA in this space.

But exactly that it is. Years ago when they came with their mini/nano/pico ITX stuff it was quite interesting for fileservers and such. Or maybe a little HTPC. The CPUs (and I have some of them) never showed much performance but barely enough for home server purposes and with 2d acceleration by the GPU also ok for DVDs and such. The problem have always been the graphic drivers - no other part of them has caused me so much headache since the old KT133 chipsets.
I had them all. The old xorg via driver (which covered hardly CLE266), I still hear Luc Verhaegens complaints, the unichrome one and the openchrome. Some did sometimes work on certain things and others not. I won't mention the binary thing since it was even more awful.
So years go by, my CLE266s are "Asbach" old and still there is no decent support. (As far as I heard the Windows drivers were of same or even inferior quality.)
So it is aging more and more and there is still no real support for it. Sometimes it feels like using a VESA driver.

I mean I really liked the idea of these C7 eden ones, the padlock stuff, the wattage is really neat for an x86 system, passively cooled and performance ... well, it could be worse.
But then when I see all the troubles I had with it... no. Not again. I'm well awara that there is hardly a perfect, flawless chipset and all but that experice with VIA isn't one that I'd recommend.

And especially that low power corner (it ain't low cost, the complete mini ITX board was about the same price like my far more powerful AMD solution), this corner was one that is now becoming populated with intel's Atom (AMD's Geode seems to be gone, though it was also an early low power soultion) and possibly nvidias ion stuff or whatever it was called. Yes. But VIA had the chance to get a good stand here already but they were so uncooperative and not willing that it seems to be too late.

[bulletxt]

Yep, I am always voting with my wallet. My next purchase will be an ATI discrete card to celebrate the 3D Open Source drivers whenever that happens.


I hope you understand that is not happening before 2011.

[chithanh]

This current situation is similar to when the Intel Poulsbo DRM was proposed for mainlining in the Linux kernel, but that ended up being rejected on the basis of the 3D driver being closed-source and the code being undocumented. VIA's code is really not any better.
The situation is very different from Intel Poulsbo. Intel was neither willing to provide open source code nor documentation. VIA on the other side has documentation (minus the pixel shader which is going to follow soon according to Harald Welte).

So as soon as the documentation is complete, the DRM could probably enter the kernel.

[Svartalf]

The situation is very different from Intel Poulsbo. Intel was neither willing to provide open source code nor documentation.

Sorry, but you have that dead wrong.

Intel was willing, but unable, to provide EITHER source or documentation for Polusbo.

They CAN'T as it's not their GPU core like all the other GMA devices.

It's Imagination Technologies' (It's a PowerVR SGX 5XX series core...) design- and they have yet to show signs of even considering the access of technical information for any purposes of anything other than a closed driver for ANY OS. Moreover, they're the ones that've bogged down releasing a blob driver for the flagship OS on one of the flagship SoC's for their chip. It's still...fun...trying to do 3D stuff on OMAP3 devices.

[Regenwald]

as i already pointed out in another thread, intel seems to open up the pulsbo driver in q4 in 2009

[chithanh]

Sorry, but you have that dead wrong.

Intel was willing, but unable, to provide EITHER source or documentation for Polusbo.Intel willingly entered into an agreement with ImgTec which forbade release of source code/docs. That amounts to the same to me. That it left their open source developers in the cold is an unfortunate side effect.

[Nille_kungen]

Well i hope that the driver is rejected from the kernel until there's full documentation.
I don't like the idea of an kernel driver that needs an binary blob for 3D.
I hope VIA one day can be as helpful as ATI against the community, but i doubt it.
Intel and ATI shows that it can be done.
So do you think the next generation of VIA graphics will be fully open source?

[Zajec]

Last time I criticized VIA for their (lack of) work on Chrome 9 I got answers from BruceChang and Jon Nettleton that I'm wrong and they just need some time. Yeah, sure. Once again we received nice "F you" :/

Check out "Chrome 9" thread in these two months:
http://wiki.openchrome.org/pipermail/openchrome-devel/2009-June/thread.html#347
http://wiki.openchrome.org/pipermail/openchrome-devel/2009-July/thread.html#385

Come on, how many ppl they need to hire to work on open source driver? I'm sure AMD provided it's not a big amount to have nice effects.

[FunkyRider]

VIA is always 'the piece of shit' IMO. I do everything I can to stop people from buying VIA products because it really sux. Now the worst has become even worse. How good could that be?

[Svartalf]

as i already pointed out in another thread, intel seems to open up the pulsbo driver in q4 in 2009

If that's the case, then there's been a shift in attitude by Imagination- which would be a rather pleasant surprise. It also translates into access to make FOSS drivers for the OMAP3 GPU since they're in the same series.

But, I'm not holding my breath on it. I'll believe that one when I see it.

[bnolsen]

VIA hasn't been seriously pursuing the gamer/enthusiast market for years; they've been setting the standards for small, low-power PC systems. Intel and Nvidia are just now getting around to producing stuff that can compete with VIA in this space.

Let me fix this for you:

...they've been setting the standards for small, low-power, HIGH PRICED PC systems...

and their market evaporated almost immediately after the atom came out.

VIA's history has been about conscienciously squandering every opportunity they ever had. This is no different.

[susikala]

Now that's a pity. I was really looking into buying the Samsung NC20... it looks like a nice little thing. :/

[Ant P.]

I wonder what's keeping VIA alive at this point? Microsoft bribes?

Their hardware hasn't been worth buying for a long time - other IGPs are already good enough for Windows users who are only interested in shiny window borders, and their track record of instability makes it unacceptable for people who want their OS to actually work.

[Ex-Cyber]

I wonder what's keeping VIA alive at this point? Microsoft bribes?Probably just coasting on past successes. They seem to be losing money, but it also seems that they've been able to more or less afford it so far. I'm no business expert, but I get the impression that they'll fold if they don't score a major success in the next few years or so. OTOH, maybe the parent company (a major Taiwanese chemical conglomerate) will decide to keep them around just for the sake of having a local vendor of PC platform chips...

[zoomblab]

I think that we should be thankful for those companies that have given open source drivers and documentation, but also respect the choise of those others that are unwilling or unable to do so. Free software is more like a GIFT than a vested right!

For the record, I own a VIA EPIA SN motherboard and I know first hand how bad their drivers are. All in all I find this "EITHER open source/documentation OR we crucify you" mentality harmful.

[Max Spain]

I wonder what's keeping VIA alive at this point? Microsoft bribes?


I have no idea of that but they did join M$ on the hardware DRM issue around the time when Vista came out.
VIA has taken a novel approach to security that deviates greatly from current DRM-driven industry trends. Instead of siding with the decidedly unpopular requests of the music and movie industry and implementing "sneakware" technology that prevents fair use copying (and even provides backdoors that can potentially allow governments and big businesses to secretly snoop the bits of your drive) VIA provides a plethora of extremely powerful hardware based security features that empower the end user, while remaining flexible enough to be used by content providers - without any sneakware.
I purchased a VIA based system back then because of this stance. These days Trusted Computing tech is inside their Nano CPU's and many of their mobos have TPM's :( It appears to me that Via believed that their M$ customers were in the majority and so they hung their other customers out dry, but I can't think of anybody who would ever want to buy a Via system to run a M$ OS (especially Vista.)

[Ex-Cyber]

I think that we should be thankful for those companies that have given open source drivers and documentation, but also respect the choise of those others that are unwilling or unable to do so. Free software is more like a GIFT than a vested right!They have the right to make it a pain in the ass for open-source operating systems to support their hardware, and I have the right to spend my money somewhere else and tell others my reasons for doing so.

[Ant P.]

Free software is more like a GIFT than a vested right!

Limited copyright is a gift. Some greedy companies see eternal copyright as their right.

[Adarion]

They have the right to make it a pain in the ass for open-source operating systems to support their hardware, and I have the right to spend my money somewhere else and tell others my reasons for doing so.

Yep. Exactly.

By the way also Windows users suffer from closed drivers, I remember the many hassels I heard people had when older Windows versions were no more supported or when and older hardware would not run in a newer Windows environment. Or if a driver was shoddy and nobody could repair it.

Iirc. a faulty printer driver (closed source) was the reason for rms to found the fsf. Nobody could fix it and the manufracturer did not want to.

[Chad Page]

I hope you understand that is not happening before 2011.
Hardly... AMD has their own people working on it - glxgears is (kinda) working in Mesa master on 4xxx's now. I think the driver'll be in good shape in Ubuntu 10.04, if not even 9.10.

[SarahKH]

I remember many months ago when VIA were releasing the documentation that I said words to the effect of: VIA will dump some/the documentation then sit back and go "why should we? Here's the documentation you do it".

I. Was. Right.

VIA's products in recent years have yet to impress me. From the mini-ITX board that died after 6 months of use (1 year + in storeage so no warranty) due to bad caps, the SATA controller that wouldn't see SATA drives through to an old super socket 7 board who's AGP port wasn't quite as up to the specifications as it ought to have been.

VIA, SiS and PC Chips (are they even still around?) wouldn't touch any of them with a barge pole personally.

[iVistux]

...have an insider giving an overview about featues and funktions that are or will be available for Chrome 9. The Webside and Mailinglist of Openchrome are not absolutly clearing the situation.

Especially Dual-Head-Funktions, 2D-Support (Video), future features like KMS and working Suspend/Resume would be interesting, cause to me decent GPU-Support is the critical argument to purchase an NC20 or not. No 3D is "ok", but the mentioned stuff should work "in the wild" (e. g. Dual-Head).

If VIA can't provide documents and/or code due to third-party-licence-problems, this is ugly but should be tolerated by us. The company should improve its communication to the community.