Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
Silicon Graphics

SGI to Dump NT Workstation Business, Move to Linux 156

Anonymous Coward writes "As part of its new restructuring SGI is spinning off its unprofitable NT workstation business and its Cray divisions. It will instead shift its Intel based products to Linux, integrating IRIX into Linux open-sourcing the merged technology, in preparation for using Intel platforms and Linux exclusively, according to Richard E. Belluzzo SGI CEO. " A lot of old news, but it's interesting to see that that they are spinning off the NT workstation business as well.
This discussion has been archived. No new comments can be posted.

SGI to Dump NT Workstation Business, Move to Linux

Comments Filter:
  • I bet this means an end to any SGI - Microsoft collaboration, look's like Open GL will be the domonant Graphics API. I guess this is good news for John Carmack :)

    I wonder how this will pan out for SGI, but it certanly bodes well for Linux users.
    "Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
  • The article says that SGI will integrate IRIX with Linux and release the result. Their changes may or may not be integrated into the kernel sources, and thus may or may not be widely adopted. I can easily see SGI supporting their own version of the kernel...bad news!

    Nevertheless, this certainly is exciting -- I always like to see this kind of thing happening.

  • And how the hell one expects to use Linux for graphics related tasks ? No fricking GL drivers, XFree is slow like dog ...
    Come on !
    NT is much better suited right know for this kind of work.

  • "The traditional supercomputer market is not really going anywhere anyway, with the trend clearly more towards commodity architecture clusters typified by Beowulf."
    These are two completely different architectures, designed for solving very different problems. Why do people think that they are always interchangeable.
  • Just because it's avaible (for both the Mac & Linux) lends little to the argument. Developers won't use it until they can depend on it being included with the OS. Unless they want to license a version from someone, in which there maybe 4 or 5 different providers who all have slightly different implentations of it, and therefore you'll end up with several different versions of software that provides nearly identical functionality.

    I'm not trying to bash anyone here, I'm just saying that we'll see a lot more 3D development once at least Redhat (which happens to seem to be the most popular distro) includes OpenGL as an option in it's installer.

    But then you'll see Apps with requirements of Redhat Linux 6.x rather than just saying that they're for "Linux". And you may have to endure downloading the software from another site est the licence includes verbage to the effect that you can't distribute the software with something that you money for.
  • Personaly, I fail to see what diffrence an OS would make as far as graphics. Remember, were talking about *non-realtime* rendering, and stuff like photoshop. The hardware would matter *a lot* more then the OS. We arn't goint to involve the OS at all (I think), unlike things like Web Server.
    "Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
  • For SGI there's only one thing that counts: selling hardware. Their IRIX workstations have long been top quality graphics workstations but now slowly IRIX is beginning to backfire. SGI does not have a large portion of the UNIX market -> nor does IRIX. Keeping an OS up to date with modern technology is hard and expensive. So if SGI changes to linux (NT fails on the modern (new :)) technology part), they gain a lot: (1) thy no longer have to maintain an entire OS (2) suddenly a lot of applications can run on their systems (3) they have a free ride on OSS, great for PR.

    They still have their hardware patents and knowledge to guarantee that they can deliver superior hardware (after all they've always targeted the high end workstation market)

    So to me it seems perfectly logical that they do this in order to survive.

    Of course it may backfire:
    Company X comes up with a nice workstation that is able to compete with SGI's stuff, Old IRIX users are pissed of because IRIX is no longer supported, ...
  • well, it's pretty unlikly that all the OSS people will stop developing Linux, but if they do SGI can afford to pay the people who worked on IRIX to work on it
    "Subtle mind control? Why do all these HTML buttons say 'Submit' ?"

  • If, as a general concensus you believe that the main downfall of SGI of late has been,

    1- Oversaturation of UNIX variants, of which they are one of the weakest
    2- Recent advances by PC technology to provide powerful visual workstations, a direct competition
    3- lack of interest in MIPS hardware, fierce competition from IBM and Sun in the ever encroached high-end server market.

    Then someone about a year ago of course would have seen a dead end road coming about and oust the CEO. A immediate move could have been to join the competition (if they are moving to NT, then we'll compete on NT.)

    It is good for them to see the other alternative. One that benifits the hardware manufacturer as well as the users. It is a gamble, to move this direction. It will pay off if other big names (IBM, HP) join in. Does it look like that might happen?

    I'm now waiting for the fall of AIX, I think it will be next. (I don't count BSDi or SCO, to me they are dead already.) Indeed, IBM may be working in that direction since the main money (support and hardware) they can still provide. They are supporting Linux on almost everything they depend on AIX for.

    I think IBM is simply bigger and it takes more time to maneuver, but they are working that direction also. Once Catia, and other big dollar Apps finaly switch over, what will there be left to stick around with AIX? Surely they see it, maybe they even want it.

  • Agreed. I smell a dead horse.

    It seems to me that SGI hasen't stuck with any product long enough to actually sell it and develope a market. It is cool to seem them jump on the Linux bandwagon. Hopefully we (the users) will see some good stuff come out of it. However, I can't help but think that SGI is hopeing that Linux will kick a bit of life back into a mostly dead company.

  • Linux is fucking at least two times slower on the desktop .

    Even if that were true, which I don't believe, at least not as a general case, what does it have to do with what we were talking about? The example was talking about IRIX, not Linux.

  • "The traditional supercomputer market is not really going anywhere anyway, with the trend clearly more towards commodity architecture clusters typified by Beowulf."

    These are two completely different architectures, designed for solving very different problems. Why do people think that they are always interchangeable.

    They are clearly different architectures. The traditional (read X/YMP series) Cray architecture machines were superfast uniprocessor machines, however many of the later Cray machines were tightly coupled multiprocessor machines. Beowulf machines are loosely coupled clusters of (usually uni but sometimes dual) processors.

    However it is obvious that most of the traditional supercomputer market has been in decline for some time, with such former vendors as Control Data dropping out entirely. Even many of the tightly-coupled multiprocessor supercomputer vendors have had serious financial problems. Cray wasn't doing well even before SGI bought them out.

    Although they require different types of programming and have their own set of strengths and weaknesses for solving certain types of problems, machines of the loosely coupled Beowulf type are clearly taking over the market from specialized supercomputer hardware.

  • First of all, if anyone takes the time to read the article instead of looking at the Anon cowards comments you know SGI is not "dumping NT" but spinning it off to a partner (my guess is Dell, because Dell was already re-selling their flat panel monitors).

    It never ceases to amaze me that slashdot, which normally has decent oversight of articles, will post the biggest piles of biased shit when it comes to anything with the word Microsoft in it. Last time I checked, the slogan was "News for Nerds", not "News for rabid Microsoft-haters". I guess the whole world is getting more tabloid by the minute.

    Secondly, if anyone has any experience with the SGI NT Workstations, you know that they were providing commodity hardware at a premium price. I'm only going to pay so much more for a cool purple box. Plus it had some weird quirks like a proprietary output for its flat panel and only one video card to support it.

    The Microsoft-haters out their can view this as problem with NT, but you would be kidding yourself. SGI is hoping the Linux hype can float their boat a little longer. Fine, whatever helps your company, but don't bash NT because SGI has problems.
  • by edgy ( 5399 )
    This is a good thing. SGI has a lot to offer to Linux, and the more companies that have a vested interest in Linux's success, the better the support we'll get for Linux.

    With the neat things like the journalled file system and other enhancements, Linux is going to be looking better and better. I think Linux finally has enough of a foothold to stay in the picture no matter how hard MS fights.

    Knock on wood.
  • My girlfriend likes my fckn at least two times
    slower on the desktop. It is to rough if I do it
    the same speed as on the bed, hard sufface and all.

    P.S. Has your mom gotten your back to school cloths
  • It's not the machine
    thats bitty and crappy.

    It's the os.
  • I suspect that a slight problem with this approach could be that on NT, many other applications are available than on Linux. At least in the heavy graphics/modeling/animation/rendering areas. This might upset people who use the Visual Workstations on NT. On the other hand, this might mean that more 3rd party people get an incentive to port their apps to Linux, which of course is always a good thing...
  • You are boring ..
  • I was specifically talking about putting that much money *on your desk* as your workstation. Do you really have a $80k PC on your desk? (which is for your personal use only, not just a server that happens to be stored on your desk.)
  • by drwiii ( 434 )
    That didn't take long at all. Bye NT.

    Don't get me wrong, it's great that SGI is adopting Linux, but I can't help wondering if they'll dump Linux when the "next big thing" hits their marketing department.

  • The company will spin off its...Windows NT Visual workstation business...

    All of this was stated in the press release yesterday. But how will SGI spin off the Visual Workstation running NT and keep the same system running Linux? It's the same computer, isn't it?

    SGI announced a few weeks ago that the Visual Workstations would come with Linux pre-installed this fall (right now it's just "Linux ready"). But the press release from yesterday (as far as I recall) didn't mention Linux on this hardware. In fact, I thought they might have meant the entire Intel line of workstations would be spun off.

    This sort of thing has cropped up before. And it has always been due to human error.

  • Well NT may not be the best at graphics, but it's got a whole lot more going for in that area than Linux does right now. We need to get some really good (stable) OpenGL drivers out. I personally cannot wait for XFree86 4.0 - my TNT2 wants to play.

  • Hopefully that shared code base won't have a heavy religious slant in the Linux direction. There are code bases out there just as good, if not better, that don't deserve to be chucked out the window just because it isn't Linux.
  • This move really scares me. Although flexibility is definitely a good thing, I am getting the impression that SGI has a short attention span. First the stupid rebranding a few months ago, now this. They need to be able to stick with a plan for more than a year. The visual workstations may not have been selling like hotcakes, but they are fairly new to the market and sport cutting edge and proprietary interfaces (cobalt graphics & pci64). Peripheral manufacturers are just now starting to catch up.

    And positioning themselves as a player in the non-existant streaming media server market seems rather ridiculous. A good server just needs great disk/memory IO, which competitors like Sun already have.

    SGI has lost its dominance in the graphics market and is now confused. Like netscape, they are forever repositioning themselves because they can't find a strategy that makes a profit. Netscape was a client company, then a server company, then a portal, then AOLs toy.

    So SGI do a good thing, embrace open software. Not only is this good for the community, it is also good for their business. IRIX systems never developed the 'network' of vendors and support that is necessary. With Linux's seeming dominance of the UN*X field (is there any UN*X platform that doesn't emulate linux binaries?) it will be much easier to integrate SGI's products into existing environments.

    I only hope that SGI can keep its focus long enough and continue its great R&D so that it can actually make an impact in the industry. At least if it does die a pitiful death its legacy will live on in the code that it opens.
  • Dumping Cray is smart just from the standpoint that SGI never seemed to be able to really capitolize on Cray. The traditional supercomputer market is not really going anywhere anyway, with the trend clearly more towards commodity architecture clusters typified by Beowulf.

    Dumping their NT sales is smart because SGI is not equipped to compete in that sort of market. They need to be in areas where people are willing to pay for quality (the high end). People who are the customers for NT in general are not interested in paying for quality. They want cheap, commodity hardware/software. They are often not willing to pay for balanced subsystems or high performance I/O. Many buy based primarily on clock-speed and whiz-bang advertising. They look at SGI's $5000 450MHz Xeon and think that some generic clone vendor's 450MHz Xeon box is comparable, just because it has the same processor. SGI just isn't the right company to make it in that sort of market. Even the established high end PC vendors (Compaq, HP, IBM, etc) have had trouble being consistantly profitable with their high end PC workstation business, and none of them really went to the extremes of breaking out of the PC design mold that SGI did.

    Unfortunately for PC-clone vendors, Microsoft has structured the market so that most of the potential profits go to Redmond, and that just doesn't leave the margins there for companies like SGI.

  • I'm afraid their reasoning goes somewhat like this: We need to cut costs. We can't afford to keep developing our own chips, so we'll let Intel do that for us. Given that we won't be using our own chips, it's pointless to keep developing our own operating system. It's cheaper to keep adding value on top of Linux, so we'll do that.

    I don't mean to question that this is a good thing for Linux and for the community. Their support for open-source software, with no fancy non-GPL licenses and no strings attached, is unequivocally better than any other Unix vendor's. Actually, I think the kind of cost-saving outsourcing opportunity that SGI sees in Linux is a really good thing for everyone concerned. Hopefully the next few years will see an explosion in this kind of cooperation and a dramatic reduction in wasteful duplication of effort, just as RMS has always predicted.

    No, I'm more concerned about the idea of blindly embracing Intel's latest product, which looks to be the kludgiest chip ever invented, with horrific cooling requirements, two silicon-wasting legacy instruction sets, and tons upon tons of cache to make up for absent registers. That's not to mention the fact that the damn thing still hasn't been released.

    It's like SGI has decided that high-end computing doesn't pay off, so it's just trying to be another PC vendor. Where is the visionary company of yore? I hope that, at the least, they will keep their position of innovation in graphics. It looks as if they're trying to do that.

    Good luck, SGI.
    Beer recipe: free! #SourceCold pints: $2 #Product

  • This recent string of announcements seems funny to me. When I think of SGI, I think of the company that brought us Open GL and high-powered graphics oriented computers. All any of these articles are talking about is servers and what operating systems they are running. Big deal. Companys that make servers are a dime a dozen. I want to know what they are going to do to make Linux a serious graphics platform. There is a lot of talk on this subject now (things that will appear), but Pixar and ILM aren't going to buy SGI Linux boxes if they can't do anything with them. Support for there proprietary graphics hardware under Linux and some serious 3D modeling software that takes advantage of it is what SGI needs, not strategy shifts and new names for there server lines.
  • How is this bad for the trial?

    Did MS have a monoply in the past? I think so. (But this isn't illegal in itself.) Did MS abuse this monoply? I think so. (This would be the illegal part.) If the trial finds them guilty it will be for actions in the past that abused a past monoply. MS can lose this monopoly, but the court can still find that they had (past tense) a monoply and abused it to gain an advatange in new markets and had anti-competitive practices.

    If you commit a crime and then become the most upstanding citizen in town, it doesn't make you any less guilty of the crime. Even if you are truely sorry and realize the evil of your ways the crime has still been committed and you are still the guilty party. Your sentenance may be reduced, but I doubt MS will get off so easy they didn't repent before they started getting investigated. If they are found guilty they were caught pretty much red-handed.

    But then again, IANAL.


  • Avid's completely pulled away from the Mac.
    Avid has since rescinded their decision to abandon on the Mac platform. Turns out they reason they had the impression there were more NT people internetested is because Avid never heard from most Mac folks about problems, upgrades and the like. The Mac folks, when hearing about Avid's decision, soon got on the horn and the keyboard, though.
  • by Anonymous Coward
    Oh, Ricky, I am so sorry for all the names I called you. When you were at HP and HP was on the verge of dumping HP-UX for NT... And then you left HP for SGI, which suddenly was about to become just an NT reseller...

    But I guess you weren't a total borg unit after all.

    Maybe you're still a trend-following clueless corporate doofus, but dammit, you're our trend-following clueless corporate doofus!
  • What would be really cool, but may not happen, is if SGI decided to open source IRIX, or contribute some of its code to Linux. IRIX has some great things in it, such as 3 shared memory standards, high-end graphics support, etc. Some of that code would be a really great thing to have in the open.

    I'm always sorry to hear that SGI is having trouble making money. They have always been a great company for one thing - ideas. They have never been afraid to cut their own path and innovate a little. Maybe they can make some extra cash from working with NVidia.

    As far as NT goes, I'm glad they decided to switch to Linux. When high-end, innovative companies choose Linux, it makes the platform look good to other such companies.

  • Wonder if they will be making their own Linux Distribution. Irix Linux style =) kinda catchy...
  • IBM and SGI are great R&D companies, but have lossy marketing and market positioning. IBM has been positioning AIX as a stable web server, but they are also jumping all over the Linux bandwagon. IBM is also helping SCO (and others) to create Monterey, the "new Unix". What is IBM trying to do here? IBM has great technology, but seems be have the big-company-syndrome where the right-hand doesn't know what the left-hand is doing/saying. IBM is just confusing its customers (and itself) by pushing so many different operating systems.
  • What they're really doing is shifting from the highly competitive desktop market to internet servers. Since Linux has always been the preferred OS for backroom web servers while NT has always been the desktop choice, they're naturally dropping NT. It's no new big news for Linux.
  • GL is hardware acelerated on the 320/540 under
    linux. However, texture mapping is still a little
    buggy. Expect to see this available to the public
    in the near future, pending the release of
    XFree86 4.0 along with the new DRI. Interestingly
    enough, HP also has hardware GL acceleration under
    linux for there visualize workstation.
  • yeah, but at least if it dies, it will go out giving lots of cool technology to linux. It could have died a lot of different ways, and a lot of great work could have been lost forever. Now, linux can assimilate the best parts of irix and add its diversity to its own. Soon galactic domination!

    p.s. Also, I am still trying to forgive them for putting the PCI slots in their VW's backwards! I still haven't gotten over that. =(
  • They're cool for desktop publishing,
    but they've long since lost any edge
    that they had in any of the standard
    graphics markets, and are (Despite
    they're current efforts) worthless
    for 3D.
  • Hello? Earth to Wanker? The 1400 is a server, not a visual workstation.
  • the 320/540 outperform the dell boxes and would
    win in price/performance if SGI hadn't made a critical error, every VPC comes with video editing hardware. Most of the workstation users simply want a CAD solution so they _may_ opt for the other box, but here is where you're seriously misinformed...
  • I'd certainly agree that there are other code bases out there with some great features. What we can hope for is to see them incoporated into Linux or some other open-source OS so everyone can benefit from them. That is why we should root for Linux over proprietary Unices. As more and more companies turn to Linux, I expect a bit of the zealotry to subside; we just need to stand fast and demand to see the source.

    Online multi-player Open role-playing game? World Forge []
  • Ummm it's not the OS per se, just the services it provides...

    The MacOS has long had a "WYSIWYG" interface along with, more recently, color management. Hense the reason that it's still the favored platform for graphic designers, even though most if not all the apps are also available on Win32.

    It does not, however, include OpenGL... Instead, it's been using Quickdraw 3d, and therefore there aren't as many High-End apps available (yes, lightwave, electric image, and formz exist, but what about the others?)

    Win NT does not have color management integrated with the OS. It does have Open GL, which makes it much easier to port Unix 3d Apps...

    Linux, does not have OpenGL. It doesn't even have a standard GUI (past X)... This makes it much more difficult to port applications, or else will create immensely bloated applications, as each vendor includes OpenGL functionality rather than using what is already available within the OS, which in Linux's case is nothing.

    Then you have Be and their "pervasive" multithreading, therefore makes it a good choice for video apps where nearly uncompressed video is streaming to and from the HD while the OS processes other tasks...

    i could go on and on, but i've got work to do! Moral to this story is that the OS is at least as important as the hardware it runs on
  • the 320/540 outperform the dell boxes and would
    win in price/performance if SGI hadn't made a critical error, every VPC comes with video editing hardware. Most of the workstation users simply want a CAD solution so they _may_ opt for the other box, but here is where you're seriously misinformed...
    In the workstation market, the cost of the computer is a very small percent of the cost of seating an engineer at a workstation. Most of the cost is in paying the salary of the engineer. So _any_ small increase in the performance of the workstation will boost the performance of their larger investment in the engineer. Pay a little more for your workstations or pay a hell of a lot more for more engineers, not a tough decision.

  • No.

    They have pci slots, but its about 6 times faster than AGP2x.. 3.2GB/sec... you use your RAM for graphics memory... Cobalt graphics... ATI is a really silly comparison..

    and as for personal experience... we have 2 540's... a gig of ram each... also have the dells with the FireGL boards... The dells choked while trying to work on our projects, but they render fine...
  • I probably couldn't make any money reselling NT licenses either - we just got one today from a large company, a laptop w/ NTWorkstation, preconfig'd w/ NetBEUI, so I deleted that and installed TCP/IP, and NBT worked ok, could access server shares, but none of the Internet stuff worked, no IE, no FTP, get "bind: invalid parameter". After diddling around for half a day upgraded to SP5 (came w/ 4) and that was the magic bullet. Jeeze, if you HAVE to spend time config'ing something to get it to work it may as well be open. Layperson end users STILL need access to someone who makes NT a career to manage it.

  • No one cares about whiz-bang boxes and graphics cards that are 10% better than ATI stock cards.

    not a very educated thing to say... the ati boards are not even in the same league as the sgi nt boxes/graphics... same with the i/o speed...

  • Maybe they could call it iLinux and sell it in several fruity flavors. Sorry, I couldn't resist.
  • regarding c)
    SGI is demonstrating a 320 with hardware accelerated GL on linux at LinuxExpo.
  • Heh, quite a coincidence, but the banner ad I have right now is for an SGI Linux box. (:

    Anyway, this is obviously really good for Linux in general - it's a huge push by a big company with a lot of resources (granted, less than they had a few years ago). I'm also glad to see NT declared "unprofitable".

    I know they're not exactly dumping Cray into /dev/null, nor giving up on MIPS and IRIX, but they're spinning them off to independant companies which is a bit sad - much as I like to see NT declared unprofitable, it's sad to see Cray and IRIX grouped there, too.

    Still... it's better than losing SGI. And maybe now there's finally a chance of getting Electropaint on Linux! for those who don't know, it's this really simple-yet-amazing OpenGL demo that came with IRIX.
  • I don't believe that Redhat 5.2 (the distro i use) included it? If it did, i'm wrong... If it didn't and the newer distro's don't either, then that's what'll hold linux back.

    Just because it's available for the OS doesn't mean that it's standard equipment, and that's what is required. Otherwise you'll see SGI with one version, Redhat with another, Caldera with a 3rd, all saying that their versions are somewhat superior to the others...
  • If SGI godes under, this could become "Let's all ditch this Linux thing before it drags us down like it did SGI!".

    This would be much more of a problem if SGI had bet on Linux when they were profitable and then went down the tubes. Since SGI is already in serious financial trouble, if they fail now, the worst you can say is that their move to Linux failed to rescue them. But you certainly can't say it was what did them in.

  • That's ridiculous. Who cares if it didn't come with Redhat 5.2? Mesa has been available for a long time, and I'm willing to bet it did come with Redhat anyway. Once we get 4.0 with the DRI then linux will really be in a good position as a platform for 3d apps.
  • Was that in jest? The VWs use SGI's Cobalt chip for some gigantic I/O performance... No AGP anywhere, although I think the SCSI card is on a PCI slot (I could be horribly wrong, too).

  • I think that Cray has missed a major income source...T-shirts, mugs, hats...I would *love* to have a Cray t-shirt, a hat, mug...anything!

  • The way I read it, they aren't ditching the Visual Workstation. They are ditching the idea of running NT on the Visual Workstation. This is a smart idea, and not just because I like Linux better than NT. The problem is that the kinds of customers who are into NT are generally not the same sorts of customers who will pay for the highest-end whizbang hardware. They want cheap Dells and Gateways. The sorts of customers who are willing to pay $5000 for a workstation that has amazingly fast busses and cards are usually the sort who don't mind using Unix. This was SGI's dumb mistake. The intersection of the set of all people who prefer NT and the set of all people who care about high bus rates and high end graphics is a very small set. They might get a few graphic designers in the movie industry and that's about it.
  • This comment is what I have been trying to drive at ever since SGI started making its linux announcements. I want softimage and the like ported to linux. I want the high end graphics stuff. While a journaling filesystem is hella cool (even thought it wont be implemented in 2.4), I want to see a graphics market ported. this is a chance for more desktop attention. Don't bitch at me about blender and gimp either ;) I love my gimp and my blender but not that many commercial shops are going to use blender right now because it means learning a whole new interface (sortof). I kinda wish blender did get more exposure though.
  • You've got the gimp
    ac3d and the blender
    to start with, and
    the povray tools.

    And Maya's already in
    the process of being ported.

    Don't have a caniption yet.
  • by Anonymous Coward
    An important thing to bear in mind when questioning why SGI picked up NT a year ago and then dropped it is that SGI had plans for the NT Visual Workstations over 2.5 years ago, well before Rick Belluzo came into the game. Belluzo saw it for the mistake it was, but was basically too late to stop it from happening. He's put a cap on it before it gets much worse. So what looks likes inconsistency in the company as a whole is really just the change of hands from a CEO (Ed) who threw really great parties on campus (everybody remember Hewey?) to a CEO (Rick) who hasn't proven himself one way or another.
  • I don't really think that SGI is betting it's future on the whim of thousand of *anonymous* contributors. It's more like, SGI see Linux as a stable system with a lot potential with thousands of seemgly *anonymous* contributors who are trying, or are willing to help anyone to tap into that potential.

    Not all contributors are *anonymous*

  • avid pulled away from the mac because it needs 4 pci slots for its hardware, and the current g3 m/b's only have 3. this is to be fixed in g4.
  • I'd agree. I think linux will be marketed at the low end. This is going to put Sun in the hot seat, since IBM will be going after them from all sides and the "low end" linux on the Merced could well be a viable competitor to the cheaper SPARC servers if the kernel dev team get SMP working in a manner half respectable.

  • hah! If they weren't so damned expensive and proprietary (half-length sdram??) they might be profitable. Sure they have expensive OpenGL graphics cards, but are they really $2000 dollar cards?

  • The article says that SGI will integrate IRIX with Linux and release the result.

    Probably a large part of that is XFS. It'll probably be a lot easier to get people to switch over if they don't have to reformat their hard drives; just install a new version of linux, set up a LILO-type configuration, and reboot. The IRIX can even remain until the user is happy with Linux.
  • on the heels of the SGI announcement, at least one "name" in the 3D Imaging world decided to hitch their wagon to SGI and Linux.

    One thing the SGI announcement does is give Linux instant cachet as a 3D rendering platform. Now all we have to do is get XFree86 4.0 out the door so that the OS matches the hype :-(.

  • WE can't get NT to run at all - never mind graphics. Good for nothing except erasing hard-drives.

    Mikkkrosoft: software that 'werks.
  • Linux + Corel = profit. Corel was losing money left and right until they latched onto Linux. Now they've had two straight quarters of profitability, and expect more.

    SGI may be hoping the same happens for them.

  • Wasn't SGI going down the drain? The only problem with Linux is it seems be to getting embraced by a lot of loser companies that couldn't make it like Corel, SGI, etc.

    Sounds like desperation more than anything else.

    More Gates roadkill
  • Thing is, SGI is a hardware company. The OS is a cost item for them in the first place, not a profit center. In that context, Linux makes a lot of sense -- think, a low cost OS that we can customize as we see fit? The thing about NT was that SGI was not allowed to customize it, thus SGI's NT boxes looked just like Compaq's NT boxes looked just like Dell's NT boxes ..... SGI could not put their "stamp" on their NT boxes to make them special, and people avoided them in droves.

    I think SGI will sell at LEAST as many Linux boxes within the next year as they sold NT boxes last year. Of course, considering the pitifully small number of NT boxes that they sold last year (38,000?), that won't be hard to do.

  • I would like to see Linux be implemented and supported on all architectures so that companies like SGI could say, "Linux is Great! And, it's even better on SGI hardware."

    Each manufacturer would contract or contribute to the common codebase so as to make sure that their hardware is well represented - then they make their profit by offering high-availability, scalability, reliability in terms of hardware.

    SGI could then say that their hardware provides specific benefits that are undeniable - integration with SGI hardware and extended XFS functions, super optimised OpenGL graphics, super high graphic bandwidth etc...

    Everybody wins because applications could be sold that run on any hardware due to Linux's common codebase. SGI applications could be produced that run on IBM RS/6000's or PA/RISC or SparC - it wont matter what architecture you have.

    I believe that there is room at the high-end; especially in clustering technologies as well as massively SMP systems. SGI should take their Cray and MIPS technologies and do some of these things. Reduce the cost per node but sell a whole lot more of them!
  • by Anonymous Coward
    IF someone from SGI is reading this or if someone "in the know" could help my dilema:

    Im a happy user of a 320 that i use to run NT apps (3D max / ACAD et al )

    I have a order pending for a 540 (that i may convert to a 320 since here no one ever saw a 540)which i am planning will be only to run linux.

    Meanwhile i saw the news about the restructuring in SGI and i have several doubts:

    a) Which is the future of the VW line? Will it be updated or is it adead line ?

    b) Will SGI (and/or the "partner" or NVIDIA ) continue to improve the excelent architecture or future VW will use crappy AGP like pecees?
    c) about Linux: could one expect to see, independently of the questions
    above, full 3D graphics on the 320/540 in the NEAR future (september)?
    d) regarding the 540 : besides the fact it can hold 4 Xeons and be a lot more expensive , is there from a performance standpoint any reason to buy a 540 over a 320 to run linux?
    e) for those on the field is it realistic to buy a SGI VW (be it 320 or 540) for Linux work?
    f) Whats the status of FULL Linux support for the VW a few months ago?

    Thanks in advance

    Best regards

    P.S.: Drop the VW line is a very dumb thing! Here in portugal SGI wont
    sell many more exclusively because it doesent have them for delivery . Besides its the NT business that is Wrong not the Workstation business ans this things properly crafted have everithing to eat SUNs lunch and be THE linux workstations and if someone from SGI is reading this belive me THERE IS A MARKET not only for servers but also for Linux Workstations!
  • Don't forget why there are going this route to begin with. SGI loves Linux, not (just) because it's the buzzword of the year, but because the system is open, and SGI has a lot more creative control over it.

    Remember that cool quote by an SGI exec: "With NT, you are really limited in terms of the value you can add to the system" (or words to that effect). That basically sums it up in a nutshell.

    (P.S.: "Value you can add" != "proprietary extensions". Check out the position paper SGI put out today. They're very gung-ho about not balkanizing Linux, just making the best damn hardware/software combination ever)
  • Corel stock raised 20% when they announced their user friendly Linux distribution.

    But SGI sounds a little desperate. Let's hope that their Linux move will be succesful. Anyway Linux is in a win-win situation. And I really like that. :-)
  • I can only imagine they're going to lean on Avid (that's who owns Softimage this week, right?) to get Softimage ported to Linux, since it's already available for Irix.

    SGI designs their hardware around the demands of this kind of program. I think all the s/w vendors know this and will port their wares, simply because they want their programs to run their best. We've all heard how easy it is to port to Linux from unix flavors. It's often just "./configure; make"

  • "The emergence of Linux as an industry standard, and the fact that Linux is better than any proprietary version of Unix, led SGI to reassess its position in the Windows NT market, Vrolyk said."

    Wow I wonder if Sun, IBM, etc. know that Linux is better.
  • No. Linux does have OpenGL.

    Also, it's not "much more difficult to port applications" because most of them are already using Unix and X, so the port is in fact easy. It's getting the "product" together -- packaging, marketing, support that takes extra effort -- for any new platform.
  • Where did you get 3.3.4 ? I might be stupid but I have had a hard time finding it. The XFree ftp directories for it were empty...coming soon or something like that... I need support for my piece o crap S3 3DTRIO card...
  • I found it on rufus and downloaded the source RPM from RH6.0 contrib RPMS. which can be found here: /SRPMS [] (the Rufus link is incorrect, as the Redhat directory structure has changed slightly.)

    If you don't want an RPM . . . use alien to convert it to a .tgz file.

  • Remember, if they make their own version of the kernel, they still have to release the source. If their version is good, people will fold their changes into the mainstream kernel. If not, who cares?

  • Wasn't that Maya port just the renderer. Anyway, that means linux needs some real good 3D apps. Something that cost $29 or $100 just can't match software that cost $30,000+. Nobody is that deluded to pay that much more. Hell, if their was some open source 3d apps, I might believe they could stand a chance against the big time 3d apps, as the price/quality ratio isn't so glaring.
  • As someone who has used Maya on both IRIX and NT, I can only agree...

  • SGI has to demonstrate that it can make money with linux. Heck, it has to demonstrate that it can just make money.

    As to supporting linux, that's great. Linux this. Linux that. Yet another marketing press release. Check company stock price. Repeat. Ho-hum.

    As to the next big thing... As long as it's unix-based or includes GNU sourcecode not only will I be happy, but I will have been involved in it. So if it's called embedix or hurd or floofloo, It doesn't really matter.

    So... to get back to the main question. How will SGI fare as a whitebox (okay, purple/grey box) manufacturer???
  • Linux still doesn't any of the management tools and high end features that AIX has and Monterrey will have.

    It may not be long until it does. As many have pointed out, they look more like they are keeping there bases covered. But then what will be the draw to AIX when linux has XFS working, and linuxconf (which could still use some lessons from SMIT) and Catia, and AFS, and... well these things aren't that far off (compated to Monterey.)

    Here's a quickie, suppose Monterey is released under some open source, but patent protected in certain sensitive areas? Then anyone can fix it, but only those who have issue to the patents can actually sell/use it. Would it be like the MPL (NPL, APL, EPL, QPL) but only useful to companies and have spin-off control?

  • Macs have pretty much a zero admin overhead.

    They're easy to get work done on - most users can switch them on & go to work. They're great for writers & graphic artists who don't want to know what the OS is doing.

    I know at of people who still have the old Mac SE, running MacWrite & a bubblejet printer. All still going strong.

    They definitely have their place in the great panoply of 'puters.

    Still, once you get a taste of recompiling a linux kernel - well there's not going back!
  • ... If IBM bought CRAY????
    -- ----------------------------------------------
    Vive le logiciel... Libre!!!
  • Er, SGI (or is it sgi) has a lot to offer anyone who asks for it. So does IBM, Corel and most other companies. I would say MS wouldn't have much to offer linux if they were to endorse it since they have a knack of shooting themseleves in the foot. Well.. IBM shouldn't contribute any propoganda as well.. we know how terrible it is at that.
  • Now all we have to do is get XFree86 4.0 out the door so that the OS matches the hype :-(.

    Last time I checked, X wasn't part of the OS :)

  • Having just upgraded to XFree 3.3.4, I didn't even have to recompile the glx module for the TNT2, and I saw a deffinate performance increase (along with the absence of a nasty memory leak that seemed to plague the Nvidia release). So what exactly you mean by "play" I don't know, Quake 1, 2, and 3 run just fine . . . and Xracer and 1600x1200 is simply amazing ;)

  • A rich bundle of SGI code entering the open source world can only be good.

    Bravo, SGI!

  • I actually thought their NT move was a good one. They were a hardware company, not an OS company and so offloading that to someone else made a lot of sense to me.

    By the same token, I think their move to Linux is a good one. They can continue to make the development investment needed to support their hardware while garnering good press from giving back to the Linux community and leveraging the growing Linux installed base, which gives them economies not offered by maintaing their own OS. At the same time, they gain more autonomy than they had under NT, since Linux gives them a bit more room to innovate than creating extentions to NT.
  • I used to love SGI, but things just have gotten sadder & sadder with them ...

    Strategy: We have over-priced servers & workstations.

    Solution: Oust the CEO, bring in Windows NT.

    Problem: NT doesn't sell & we still have IRIX servers.

    Solution: Bring in Linux, drop MIPS, and use Merced.

    Doesn't this seem like a "XXX will save us!" panacea strategy? I'm really glad this means one less UNIX thread, and I hope some of IRIX's innovations are incorporated into Linux (such as their accelerated X server). I really do want to see kick-ass Linux FX workstations too.

    BUT, Linux is just an operating system. It can't cure a sick company by itself. Execution is what counts.
  • Linux has been enjoying some very large corporate migration announcements in the last few days - Corel and SGI, to be prescise. Methinks this does not bode well for the DOJ vs. MS trial, at least as far as the DOJ imposing fines/restrictions upon the BFR (beast from redmond).

    However, I do think that it's much better for Linux to win than for Microsoft to lose. If the companies fleeing to Linux for help can actually work with the Open Source community, contributing their efforts and their publicity, then we might have a winner on our hands.

    Of course, as it has been pointed out before, as soon as Linux hits it big, most developers will leave for the "next big thing". Oh well. That's my opinion only.

  • Actually NT is really lousy for graphics, unfortunately a lot of good graphics software runs on NT. That software would be better suited to a good OS, however.

  • It doesn't say ditching everything for Intel/Linux! "SGI will continue to introduce new MIPS processors into 2002, as previously planned, Vrolyk said...Going forward, SGI will shift its focus to the Linux operating systems for Intel-based platforms, while maintaining its investment in and support for MIPS-based systems into 2002, Vrolyk said...SGI will integrate Irix into the Linux environment"

    That's the somewhat unclear part. MIPS will stick around, and SGI/MIPS hardware is something that would be a very sad thing to loose. Putting IRIX stuff into Linux is one thing, but my question is how willing are they to replace IRIX on MIPS with Linux?

    When pondering a SGI purchace on the low end of the MIPS spectrum (from old Indy's for $600 to O2's that can go over $10,000), it would be nice to have that as an additional choice to x86 and G3. But, the reason I personally won't do it is because of IRIX... I could consider some of the used SGI/MIPS hardware from a financial point, but affording to keep up with IRIX is what holds most people back. SGI has been fairly supportive about getting Linux on the Indys...and I hope that continues.

    GNU/GPL OS like Linux means you get the choice to put in some time to keep things current rather than putting money into OS subscriptions for something commercial like IRIX. That would be a new market segment for SGI's hardware, possably benifiting them greatly. But I don't see SGI making it very clear that they fully intend on helping Linux get up to speed and scaliable for some of the higher end hardware, and until they do that, I don't think they will let go of IRIX completely, only allow Linux to use parts of it so that it can be more compatiable with thier higher end hardware.

    I'm sorry I am not as excited as everyone else about SGI doing Intel/Linux, to me, it's sorta just another Linux Hardware Vendor (which is good considering it _is_ SGI, and they are doing cool things even with Intel based hardware, but not overly thrilling). What I would like to see is a clearer stance on the scale up of Linux to MIPS, and how they might be willing to support Linux projects to help make Linux more ready for >4 CPU's, MIPS hardware, etc...

  • I wish SGI the best of luck, but I'm really not sure it's enough to keep them afloat. Linux is by no means a magic bullet.


    I really do hope I'm wrong about their ability to survive, though. It may be stupid, but the way most people (particularly management) would see Linux's inability to save a dying company as a general indictment of its commercial viability.

    If SGI godes under, this could become "Let's all ditch this Linux thing before it drags us down like it did SGI!".

    Maybe I'm just being morbid.

    On second thought, all SGI needs to do is bring back the Cube. That'd make everything ok again.
  • dear me... I even used preview...


  • Good luck SGI!

    If anyone here uses the STL under Linux you should be rooting SGI on too!

  • Don't get me wrong, I'm happy to see such a large company getting involved in Linux and contributing to its development. But don't you think that it looks like sinking SGI is looking for anything to keep it afloat. It'll be
    painful to see if this involvement will not bring a sufficient financial return for SGI.

    What if SGI will not be able to hang around before its Linux business takes off? That could be easily used as an argument in "Open Source/Linux oriented business model" debate.

    Of course, I'm have no formal business education and talk here out of my ass, but I'd rather see them stepping slowly into the water than taking a fast dive.
  • If those IRIX people have stuck around....

    It's a slap in the face when your work
    is being replaced by externaly developed
    software because its 'better'.
  • I saw one of the new G3 powermacs at CompUSA today while I was picking up a new modem. I played with it a bit, played the future cop demo, started up some other app, I forget what, got the message: "there is not enough memory to run . Try closing the application "Hoyle's Casino" first."

    This was an app that had been in the background the whole time I was running. I see they still can't page.

    I walked away, saying "yep, still sucks"
  • You are obviously correct that XFree86 is not part of the kernel. But, for what must be at least 90% of cases of the use of the term, XFree86 *is* part of the OS.
  • While this applies specifically to Intel-based SGI systems -- not the big NUMA servers and MIPS workstatsions -- it may represent an incremental strategy to move completely to Linux over the next 5-10 years. SGI seems to be a company that could really benefit from moving completely to Linux. Unfortunately, they are going to be stuck developing and supporting Irix for a long time to come. I would be surprised if they could move the high end to Linux in less than 5 or 6 years.

Nothing succeeds like the appearance of success. -- Christopher Lascl