Linux Support Fades For 3Dfx Voodoo, Rage 128, VIA 330
An anonymous reader writes "The developers behind the Mesa 3D graphics library, which provides the default graphics driver support for most hardware on Linux (and BSD/Solaris), has ended their support for older hardware. Being removed from Mesa (and therefore versions of Linux distributions) is support for hardware like the 3Dfx Voodoo, Intel i810, ATI Rage, and S3 Savage graphics processors. Also drivers being dropped were for Matrox and VIA graphics. Mesa developers also decided it's time to end support for the BeOS operating system. Dropping this code lowered the developers' responsibility by some 100k L.O.C., so maybe we will see GL3 support and OpenCL in Linux a bit sooner."
A fork for old machines (Score:3, Interesting)
Sounds like it's time for a legacy fork for old machines. Or maybe just keeping old versions alive, the way Linux distros do with other libraries.
-- hendrik
Re: (Score:2)
That would be nice. This hardware is still common, easily available, and has life in it yet if there is software support. It doesn't take much to run Open Arena.
Re: (Score:3)
OTOH, if you're still running a 3DFX VooDoo 2, you're probably not using it for gaming. It still works in VESA modes, and still works as a video card for 2D applications, it's just 3D accelerated modes that won't be supported any more. I have a server that still has an ATi Mach32 in it, and it works no problem, even though it's a much older card than the lot that's being dropped now.
Re: (Score:2, Informative)
The Voodoo 2 was exclusively for 3D acceleration and had to be used in conjunction with a separate 2D graphics card. It's only use was gaming/3D rendering so it most defiantly won't work as a video card for 2D applications.
Re: (Score:2)
Re:A fork for old machines (Score:5, Informative)
Not sure that makes sense...from the article: "Code that was mostly unmaintained and didn't receive new feature support work in years." The volunteers already quit working on it years ago: this is just being honest about it.
Want to keep using the hardware? Just keep using the 7.11 release.
Re:A fork for old machines (Score:5, Insightful)
Code that was mostly unmaintained and didn't receive new feature support work in years.
Code that works doesn't need new features.
Re:A fork for old machines (Score:4, Informative)
True, but if code it depends on changes then it needs maintenance.
Re:A fork for old machines (Score:4, Insightful)
Code that works doesn't need new features.
Code that worked a few years ago is bitrotten now when the rest of the codebase has received numerous other modifications.
Re: (Score:2)
Want to keep using the hardware? Just keep using the 7.11 release.
I hear that release comes with a free Slurpee.
Re: (Score:2)
You know you're wrong.
Re: (Score:2)
Won't happen. Too much trouble for the distros. You'll almost certainly just have to stick with old distros or ones that are specifically for old hardware.
Re:A fork for old machines (Score:5, Insightful)
I agree, but I'd love to see someone (with time, experience, and more knowledge than I) take it a step further: A Linux distro to work on ancient machines, with the latest feasible versions of software.
When I volunteered in Africa in 2009, one of my projects was to set up a computer lab, populated with donated machines. These computers were old. The newest one was manufactured in 2003. The oldest was 1997. I ended up installing Ubuntu and Edubuntu, then stripped down the core as much as I could while still keeping things clean. The machines still take several minutes to boot fully.
What I'd love would be a distro designed for just such situations. On install, it would determine what kind of hardware you have available, and only install things that will work well. Support for really old hardware would be patched in for the distro, probably with only major bugs receiving repair attention. If a package isn't likely to run well on your system, it will alert you before installing.
need a lot more specifics (Score:2)
was the problem with the new software that it used too much RAM and CPU?
or was the problem that it didnt support old hardware graphics drivers?
those are two massively different problems.
Re: (Score:2)
Re: (Score:2)
Ubuntu was a poor choice of distribution; it wants to install a default desktop and to simplify installation by making assumptions (including assuming you have a relatively modern machine). Debian would've been a better choice. You could've started with a very stripped down package selection (including passing on having an X server) without much difficulty.
Re: (Score:2)
Re: (Score:3)
Peppermint or Puppy would also have been good choices. They aim (especially Puppy) to be a simple, but very lightweight distros. Puppy feels a little foreign because of the WM it uses, but Peppermint feels natural to this Gnome user even tho its not using Gnome. Everything feels intuitive, and very fast.
Re: (Score:2)
I ended up installing Ubuntu and Edubuntu, then stripped down the core as much as I could while still keeping things clean. The machines still take several minutes to boot fully.
In my experience, it's not the speed of the processor that makes an old machine feel old. It's more a combination of not enough RAM and/or slow disk. Beef those up and do a minimal install, then apt-get install everything you want. Regardless of cpu (within reason :-), it'll be more than capable of keeping up.
I just abandoned an old Sun U30 (ca. '99), not because it wasn't fast enough, but because it weighed a ton. It was perfectly capable of running Linux and OpenBSD.
As for forks for old machines, there
Re: (Score:2)
Re:A fork for old machines (Score:4, Funny)
Rant please :)
Re: (Score:3)
Ridiculous. A computer from '06 can and will run any distribution released today.
Re: (Score:2)
For REALLY old machines (2003 and prior), Puppy linux works wonders. Entire distro is under 100m, and the OS boots entirely to RAM even on a machine with under 100MB of ram.
Someone tasked me with getting a 98-2000 (i think) laptop set up and useable, so I put puppy on there and installed Opera browser (with is also wonderful for such old, crappy machines). The thing worked pretty darn good for 96MB of ram and a Pentium 2 (or whever awful processor it had). The alternative, of course-- Windows ME (what it
Re: (Score:3)
Normally I run fedora, but when I was turning a 90s comp into a router a couple years ago I dropped a new-ish debian on it, and everything worked just great. It's a fake problem. The problem is choosing distros like Ubuntu when you should be choosing a more... server oriented distro for old comps.
Re: (Score:3)
The cost you mention do not represent the actual costs of shipping stuff. Shipping a container which is not in a hurry is dirt cheap.
You could have deducted that yourself. How can a new system be shipped to Lagos from (lets say) China and cost $577. And then you mention shipping computers from the US to Lagos which would be twice (or more) as expensive (without the price of the hardware).
UPS is ripping you off.
Re:A fork for old machines (Score:4, Informative)
There's two ways to do business in Africa. First, there's the Western way: Employees do their job honestly, get paid very well for it, and get fired if they're corrupt. UPS probably operates like this.
Then there's the local way. What you do is fill a shipping container with equipment, then bribe an official, say, $200 on the condition that it arrives safely. Of course, attempting to bribe an official is illegal, but so is aiding theft. For a country where the average monthly wage is $40, that's a big bribe, and it gets the job done. Customs officials approve the shipment quickly (because they'll be willing to help a local, especially if they belong to the same ancient tribe), local truckers can be haggled down to shipping at reasonable rates, and the destination is miraculously free of thieves. Once the job's done, you pay off the bribe and get on with the next bit of business.
Or so I've heard, at least, from a guy who worked in shipping mining equipment.
Re: (Score:2)
Sounds like it's time for a legacy fork for old machines. Or maybe just keeping old versions alive, the way Linux distros do with other libraries.
Or you can just run Debian Stable
Re: (Score:3, Insightful)
...And? (Score:3)
This is news? Trimming out old cruft from a source code tree isn't a big deal.
Re: (Score:2)
Seriously. If your using 13-year old cards in your system, you're probably not running the latest software anyway.
And, since it's open source - you're free to keep compiling in support on your own. It's not like it's Windows where you would be SOL when a manufacturer doesn't release a driver compiled for the latest x-bit version processor.
Re:...And? (Score:5, Interesting)
Re: (Score:2)
How is Doom 3 "latest software"? (Score:2)
It was released in 2004, 7 years ago. If that's the most recent thing you can show running (never mind the really nasty hacks necessary to make it run) well that just furthers his point: You aren't running modern software.
Re: (Score:3)
Not running the latest software? Doom 3 running on a Voodoo 2 [3dfxzone.it] ;)
Doom 3 was released 6 years ago. That hardly qualifies as "latest software".
Is Doom 4 out yet?
Re: (Score:2)
In today's networked world there unfortunately isn't much choice but to stay on the software upgrade treadmill. I could imagine using an unpatched box as an X terminal with no direct link to the internet (not even running a browser locally), but nothing more.
As for maintaining your own graphics driver, keeping up with the evolution of Xorg and OpenGL and the kernel, good luck with that. (W
Re: (Score:2)
In today's networked world there unfortunately isn't much choice but to stay on the software upgrade treadmill/quote?
Explain to me why Win2k with Opera 10.5 would be insufficient to browse the web?
Re: (Score:3, Insightful)
It would become part of a botnet within hours, that's why. Once your OS and web browser stop getting security updates, the clock starts ticking on the bad guys finding some unpatched vulnerability and your wandering into some trap they've set for you on the net.
Re: (Score:2)
I actually have a high-end server, built in the last year, that's using an ATi Mach32 video card. That's a 20-year old video card in a system that's less than a year old. It works great in text mode... if I *wanted* to run it in graphics mode/X I would be able to, as the card supports VESA. You don't need a dedicated driver to use VESA, as VESA is itself a standardised driver that is not being removed from Linux any time soon.
Re: (Score:2)
Wrong. I've got a Pentium III machine with a R128, with the latest version of gentoo installed. Which of course, as you pointed out, will mean I just compile in support, it's no big deal, but people are still running old hardware with modern distributions.
Re: (Score:2)
If your using 13-year old cards in your system, you're probably not running the latest software anyway.
The complete wrongness of your absolutely foolish assertion is demonstrated by the fact that I run up-to-date Lubuntu on an 8 year old Sharp Actius AV18 (laptop w/ 768MB RAM and ProSavage8 video). Chrome and AbiWord are very responsive.
Re: (Score:3)
Well, to be fair, how often does a project cut 100k Libraries of Congress out of their product.
Re: (Score:2)
I just bought a brand new Matrox graphics card last month to handle 4-8 simultaneous displays.
I need to see if the support is provided by the MGA driver they are now dropping, or there is a newer driver with a different name.
Re: (Score:3)
Re: (Score:2)
Re: (Score:3)
My guess is that they plan to make some changes in the near future that will break it, and nobody cares enough to update it.
Indeed. Old code can be allowed to "live and let live" until such time as architecture, compiler, etc. changes make it uncompilable. Then you've either got to figure out what the hell seven-years-ago guy was doing or cut it out. And if you can live without it (mostly) then the choice is pretty clear.
Re:...And? (Score:5, Insightful)
Because if it hasn't been changed in 7 years, then chances are that nobody really knows the code. Consequently nobody is checking it nor is anybody likely to be paying attention to any breakage which might occur if they change the infrastructure and ultimately it's one more area in which a security vulnerability could pop into existence when somebody changes some other code.
Having essentially dead code in a project isn't a wise idea in most cases. But beyond that it's extra bandwidth that's not necessary for nearly everybody.
Re: (Score:2)
Since the drivers for all of that old hardware were unmaintained, they were hurting anyone who needed to change the driver interface for any reason, because it would
Re: (Score:2)
Re: (Score:2)
According to the dev on the mailing list, it was a pain to change anything.
Re:...And? (Score:5, Interesting)
Re: (Score:2)
I know that this case is special because it is a FLOSS project and very few of the users directly support it, but:
The developer needs to fulfil the requirements of the client. It is not the developer's place to dictate the requirements to the client.
Now I currently run modern hardware, but I've been stuck in that legacy hardware ghetto before. And it ain't pretty.
Re: (Score:3, Funny)
> The developer needs to fulfil the requirements of the client. It is not the developer's place to dictate the requirements to the client.
Here's your refund.
Re: (Score:2)
And the Developer has to weigh which client needs which features. If a feature is needed by 2000 clients and another feature only by 5 clients, then it makes sense to prioritize the feature needed by the 2000 clients.
Also, if one client just demands "gimme feature", and it would take 100 hours to implement, and another client says "I would like a feature, and I have prepared this pach here that works with the current development tree, and am willing to check each new version if it breaks anything" then the
Re:...And? (Score:4, Insightful)
Say we want to further add some animation support to the video renderer for use on the OSD, that code needs to be written for each of Xshm, Xv, OpenGL, VDPAU, VAAPI, XvMC, and PVR-350 framebuffer output. In addition, that's primarily the work of one single guy.
And people wonder why some F/OSS projects have a slow rate of development.
Re: (Score:2)
we just can't manage to get you people to migrate to more modern hardware.
In my situation at least, that unmodern hardware:
a) works,
b) is paid for,
c) is stable, and
d) most importantly, is embedded into a laptop motherboard.
The issue is that many, many wives look askance upon buying new laptops and throwing out perfectly functioning kit just because newer s/w doesn't support it.
Re: (Score:2)
Oh, wait... 3D support.
Never mind.
Re: (Score:2)
Right, it's not all graphical support they are removing, only 3D support. Surely anyone actually trying to do anything 3D has already upgraded to newer hardware, or can simply stick with the old software.
It's a similar situation in MythTV. Xv is still supported for software decoding, but XvMC and PVR-350 were (extremely limiting) hardware acceleration APIs. They only work with MPEG2, and the only people who would care to use them are those still running decade old processors that sufficiently powerful fo
Re: (Score:2)
Why is it necessarily about money? If the hardware works, why should it be scrapped prematurely?
Do these $20 video cards you speak of get fabricated at no cost to the environment and have huge
energy savings over the older "underpowered" options?
Re: (Score:2)
really how well does your "old hardware" like 480P h264 level 4.1 high profile video? I'm not really sure how you make use of a voodoo3 in a mythTV setup.
Re: (Score:2)
Eh? You greatly underestimate the power of older hardware, it's not difficult for a pentium III era machine to do 480p, with a decent video card, you might even be able to squeeze it up to 720 or 1080p (with vdpau of course [on a pci graphics card]).
Re: (Score:2)
> really how well does your "old hardware" like 480P h264 level 4.1 high profile video?
My original MythTV setup 5 years ago handled it just fine actually. If not for BluRay and the HD-PVR I would still be happily using such old hardware.
famous last words of a programmer (Score:5, Funny)
"Seems like it should require almost no effort."
Re: (Score:3)
It seems that way, but that is an excellent way of killing off a product. There may be a few dozen folks still using those old cards, but you can get a much newer card with a bunch off new features that will be supported for years to come for less than $10, probably including shipping.
Supporting a bunch of products which were obsolete years ago without adequate user base to ensure that the developers can properly support it can damage projects a lot more quickly than dropping support for long obsolete hardw
Nooo! (Score:3)
Matrox!?!
I still use this card for dual- head support on my P100.
Maybe the resources freed by the team can be used in providing support for Elite/Impact framebuffers on classic SGI Indigo?
I will consider that an exchange worth making.
Re: (Score:2)
Re: (Score:3)
Mesa never provided support for Matrox' P-series cards anyway, so nothing will change in your case.
Re: (Score:2)
Re: (Score:3)
Mesa never provided support for Matrox' P-series cards anyway, so nothing will change in your case.
Wait... are you trying to imitate Jar-Jar Binks?
Only 3D (Score:5, Informative)
Xorg support for these cards isn't going away anytime soon though.
Interesting (Score:2)
Mesa developers also decided it's time to end support for the BeOS operating system.
One of Adam Jackson's fixes to X was "glx: unifdef BEOS_THREADS"
Doesn't affect me (Score:2)
Nope, not using any machine with such graphics cards anymore.
I'm replaying some games from that era though. With wine and my modern NVidia card. These games work better in wine than Win7 can run them, if I have to believe the info about these games online.
OpenCL? (Score:3)
OpenCL in Linux
I do believe that support for OpenCL in Linux is in the best interest of the GPGPU manufacturers(AMD, nVidia). Because Linux based HPC systems dominate the market and Windows ain't going to unseat Linux anytime soon. Thus you might not have all the features of the 3D stack, however OpenCL is definitely something fully implemented by AMD and nVidia.
Re: (Score:3)
doesn't anyone pay for electricity? (Score:2)
it's cheaper to just buy a new machine to use for a home server than pay for the electricity hogged up by old hardware
Re: (Score:2)
Why would you need 3D graphics support for a home server?
Then again, why would you use an OS that required X at all for a home server? Just avoid the 'Windows Tailpipe Fume Chasing' options that insist that configuration has to be done using X11. NetBSD is a good option, for instance.
Re: (Score:2)
Just avoid the 'Windows Tailpipe Fume Chasing' options that insist that configuration has to be done using X11.
Generally I agree with the idea that bash is better (I wince when IBM devs tell me I need to fire up the AIX GUI to run configuration; I'll take smitty, thanks); but there ARE things that are done far better through a GUI, or are much easier at 2 in the morning when you just for example want to create a new user in Active directory and you dont want to have to type out the full AD path to the OU where you want the user.
Whats that old saying about hammers and seeing everything as a nail?
Re:doesn't anyone pay for electricity? (Score:4, Insightful)
My home server, built in 1999, runs at a maximum of 80W. That means it takes 2 kW (about $0.20) per day, at most. For about $500 I could build a machine [kampmeier.com] that draws 20W, for a monetary savings of about $0.15 per day. In about 10 years, I could break even on what I spent on the new server, but by then, the hardware would be 10 years old again. What I have serves my purposes.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Yes, but it's absurd to spend $500 on that. VIA is an expensive brand, and if a '99 drive is enough for you, why go for an SSD?
This /. thread [slashdot.org] has some great suggestions, many of them much cheaper than that.
Oh, and have you accounted for electricity price hikes? I don't know about yours, but here it climbs steadily over the years.
Re: (Score:2)
If you argue in ecological terms, don't forget the impact of new hardware: get raw materials out of earth - ship them to production - produce parts - ship parts to assembly - ship machine to customer. This offsets the ecological "break even point" quite some time, in favor of using old hardware longer.
Re: (Score:2)
Re: (Score:2)
Except they're dropping support for VIA Unichrome, such as the low power VIA EPIA series. My M10000 only uses 10W of power.
Re: (Score:2)
Tell that people who run non-x86 architectures (like, say, SPARC, PowerPC, etc...) for all kinds of reasons. And before you ask why we don't just migrate to x86, remember that quality OSS software also depends on us testing it on non mainstream arches, uncovering a lot of obscure bugs that you x86-only guys won't trigger at all, but that are still there, lurking.
Re: (Score:2)
Mozilla syndrome? (Score:2, Flamebait)
This does not seem like a healthy trend.
Re: (Score:2)
Re: (Score:2)
Removing unencumbered open code/GUI/interface etc. which is known to work for no good reason is unhealthy.
No, it's not "known to work". That's the problem. No one ever tests with those drivers anymore, they break frequently, and no one fixes them. That's why they were removed. If anyone cares enough to fix and maintain them instead of just complaining on Slashdot, they will be added back to the tree.
Re: (Score:2)
What the heck is Mozilla syndrome? What does Mozillas new software development model even have to do with a decision in Linux not to support ancient hardware?
Already broken (Score:3, Informative)
It was also said that if someone comes along who is actually interested in maintaining one of the removed drivers, that the driver would be restored to the source tree.
Nothing to see here, move along.
Awe... memories (Score:2)
I remember buying my first "real" 3d card. It was an ATI Rage Fury 128. It had a whopping 32 megs of memory. And shitty ass drivers. Good times, good times...
Re: (Score:2)
I don't even know how much memory the thing had, I think maybe 2MB, which was a lot at the time. And now, I feel old.
Re: (Score:2)
At the time I thuoght it would be cool because you were supposed to be able to hook up sega saturn controllers to it, but it didn't actually play sega saturn games. It was supposed to play special computer versions of saturn games, but the only ones they ever came out with were the ones that came with the card.
Re: (Score:2)
Re: (Score:2)
I had some sort of S3 card too, but since it couldn't really do 3d, I didn't count it as a "real" 3d card. I was really pissed because I was going back and forth between that crappy card and an original 3DFX monster add-on card. I chose the wrong one. The original monster could actually accelerate OpenGL games such as Quake.
Good. (Score:2)
I've thought for some time that the "you can run it on ancient hardware that Windows doesn't support" is a terrible argument in favour of Linux (and F/OSS in general) for a couple of reasons:
1. Software that needs to care about the hardware (such as the Linux kernel or the Mesa library) is not exempt from this being a fast-moving industry, and updating drivers is not without cost in terms of effort. Effort that could be better spent elsewhere.
2. I think it hurts credibility to say "Support for more hardware
Why dump the i810 so early? (Score:2)
Re: (Score:2)
Graphic cards have been through a big revolution. These cards may not be that old, but technologically they are ancient.
Re: (Score:2)
Re: (Score:2)
Same here, but on UltraSPARC IIIi based SunBlade workstations running X on a R128 under Solaris 10, Debian Linux/SPARC 6.0.1a, FreeBSD 8.2-STABLE, and OpenBSD 4.9. It's the only card I was able to find with embedded FCode (needed by the SUN firmware) that runs perfectly on this architecture. It's a shame to see MESA guys so trigger happy dropping support for chips that are still running perfectly, albeit somewhat slowl