SGI Files Chapter 11 Bankruptcy

audi100quattro writes "The WSJ has a story about SGI filing for bankruptcy, but the SGI Investor's Relation page doesn't say anything." Nothing else really known at this point, but this is not unexpected.

Comments

Terribly sad

[mccalli]

Got to say that I find this terribly sad. When I started in computing, SGI used to be some magical company that I aspired to touching the hem of - sort of how Pixar is viewed today, although obviously without the narrative bit.

I know it was inevitable. I know the economics. I know various other things but still...still...it's a sad, sad day.

Cheers,
Ian

Re:Terribly sad

[WinterSolstice]

I know how you feel - it killed me when DEC was sold off. One of the best and brightest, IMHO.

Ok - time for a bit of a sad old-timer rant (feel free to skip if you think computers always came with Windows)

<rant>
I really miss the magic that was there in some of those old companies - DEC, SGI, H-P... back when IBM was the big enemy and the biggest thrill I had was reading some new press release and thinking of ways to really do something cool with it. I remember looking at the camera on the old SGI screens and wondering if Jetson style video-phones were right around the corner. I remember running a lab of Indy workstations and feeling like I had the monopoly on "cool". Back when Windows still needed Trumpet WinSock and I was playing MUDs halfway across the country on an AlphaStation.
I've never seen a documentation system as nice as "help" before or since. Compilers that took *any* major language and optimized it really well. A database (RDB) that ran so well that when we ported it to Sun it took 5 times the hardware dollars to make it work. Oracle doesn't hold a candle to it...
How about real clustering? How about a software company that makes defacto standards so effective EVERYONE uses them (like OpenGL or GLUT?)
Why is it that things like "external processors", "clustering", and "grid computing", keep getting touted as though they were new? Do any of these self-proclaimed Unix gurus even *know* why tty is called that?
For all the people who think Microsoft invented BASIC - for people who don't know that edit/tpu is the answer to the question of "vi or emacs" - and for those who have never had a RACF account; I pity you. You missed out on some of the really cool parts of the computer age. Heck, I bet a lot of the younger people on here never even coded stuff for GLIDE... and that was a *PC* level tech (and a nice one!).

I am saddened by the demise of the "science" part of computer science. In this era is there still room for wonder? As much as I delight in the cross compatibility and functionality of the new computers, I am saddened more by the lack of people who truly appreciate how we got them. It's probably the same feeling that the last steam train engineers felt as diesel engines took over - or perhaps the feeling modern diesel engineers feel at the trucks and planes that have largely replaced them.

Oh well. We've all had this discussion before, and I guess I'm just getting too old. At least one benefit of all that is having two VNC sessions open to WinXP and 5 terminals open to my Sun servers on my MBP with the full OpenGL desktop.

</rant>
-WS

Re:Does anyone still use the SGI workstations anym

[archen]

When your core business is high end and you push products through lots of R&D then have your market sapped by commodity products, its easy to overstep your budget and not adjust the business quickly enough. Even worse of course if the people in charge don't see the train comming down the tracks for a long time, which is what often happens in bigger businesses. SGI is also in a more vulnerable position than say Sun because Sun can deploy a server that is expected to stay put (and need support) for many years - hell even SCO is hanging on this way! The graphics industry is constantly in the push of new and faster, so we're seeing a "Unix" company demise in accelerated time.

Something died inside of us all...

[MindPrison]

I remember fondly my first encounter with 3D graphics, from the TRON movies, man - that was many years ago, the SGI computers was the no.1 on my wishlist as a kid - but a machine like that where WAY too expensive, and thats where the Commodore Amiga came and stole our hearts, all of a sudden - 3D became affordable, SGI did'nt belive in "3D-for-everyone" and I believe that would be the main reason for their demise.

You've got to put your belief in the little guy on the street if you want to survive, being boss - playing big, with the big - will only work until the rest of us grow up. And we did, but SGI didn't invest in our future together, if they did - we would have embraced them without as much as a seconds hesitation, but if you keep selling to the elite party (those with WAY too much money) you're out of tune with the development.


(For those too thick to read between the lines - it simply ment, they didn't follow the times)

Re:Something died inside of us all...

[Greyfox]

Computing becoming more affordable made for leaner times for a lot of those high-end workstation vendors. They were very specialized and couldn't innovate as fast as the industry as a whole. And perhaps they were resting on their laurels and didn't realize the danger until it was too late. I know the attitude at IBM at least up until the mid to late 90's was that PCs were toys and if you wanted real computing you shelled out big bucks for big iron.

When the 386 appeared on the scene and it became feasible to run a multitasking OS on a PC, the market for high end workstations dried up almost overnight. The articles hyping the processor and claiming "mainframe performance on a desktop machine!" were believed by managers world wide. Why shell out high 5 digits for a high end workstation when you could drop a couple grand and get the same performance. Interestingly enough I saw the "mainframe performance in a desktop PC" claim made for the 386, the 486 and the pentium in turn. And managers believed it each time.

Anyway the stage was set. The last act from SGI that I paid attention to was the appearance of one of their sales guys at a Linuxworld a few years ago (Must have been 98 or 99 I think.) He laid out the plan for SGI's recovery, and it involved branching into some areas of the industry that were dominated by IBM, Sun and StorageTek. I could have pointed out that they were fighting an uphill battle on that turf and that I didn't see enough of a value add from SGI to draw the customers, but it wouldn't have changed anything and I didn't want to be mean to the poor guy.

I don't know that SGI could have survived for much longer on high-end graphical workstations even if they'd stayed focussed on what they were good at. They just couldn't keep up with cheap render farms and Moore's law. Their best bet probably would have been to file patents like crazy and force every graphics card company on the planet to license their stuff, but they just weren't evil enough to go that route.

Re:The death of SGI

[ekimminau]

SGI began its rapid decline the moment the announced the merger with Cray. As the stodgy crew of maanagers went on the land grab trying to justify their existence in their "new " company, it drove out many of the long hair, fast and loose crowd of exceptional engineers who believed SGI was a magical place.

SGI truly was a magical place to be. Not only the "Its Not just a job, Its a wardrobe" pens, frisbees, t-shirts for every new product, boxer shorts, key chains, and all the other swag SGI marketing was famous for. The "O" series of products, led by the Indigo2 Max-Impact were revolutionary products. Massively fast backplanes that still exceed the performance of all but a limite few systems, incredibly fast graphics sub systems with fill rates that still can't be achieved on lowly PC gear (they just can't push the bits fast enough).

In addition, SGI truly owned the internet space, well before Sun and then gave it away once Sun started the "dot in dot.com" marketing campaign. They had the NetScape server, free, included with the IRIX OS, on every server with a full HTML configuration interface in an age where most other companies still didn't have an officially supported HTTPD for their platform. They also included Indigo Magic, the FIRST full GUI HTML editor, again, free with the OS, as well as a full GUI VRML editor, and so on.

I truly weep for the company SGI used to be. It was the best job I ever had and the one I wish had never ended.

SGI Workstations

[10Ghz]

I hope that someone will buy their MIPS-based workstation-business. What I would like to see is for someone to take that business, beef it up a bit, and port the whole lineup to Linux. I would say that there would be a sizeable market for quality MIPS-workstations that run Linux.

How about.... HyperTransport-links between CPU's, integrated mem-controllers, on-die L2-caches, HTX-expansion, multicore, multi-CPU-setups. All this, and running Linux. Hell, those changes alone would give us a nice boost, even if the CPU-core (R16000A IIRC) itself stayed relatively same.

Re:The death of SGI

[daviddennis]

I never worked for SGI, but I loved the spirit of the company and their products.

I think (and thought at the time) they should have focused on a cheaper version of their products and tried to be an Apple alternative. They had the best OS out there until MacOS X came up, and it took a long time for MacOS X to work as well as Irix did. Most people aware of the company had very warm feelings about SGI products and the OS and I think they could have used that.

I reluctantly wound up switching from SGI hardware (used Indigo2s could be had for reasonable prices) to Macs about when MacOS X came out.

XFS

[ex-geek]

My question is; where they contributing anything new to the maket recently

The XFS filesystem
I'm using this on a couple of machines. I sure hope that somebody will continue to maintain it.

This bankrupcy doesn't surprise me at all. I saw this coming for more than five years. But I remember having arguments with SGI fans who tried to defend the Indefensible.

Nice systems, but the company was a pain to deal/w

[timepilot]

I bought a few of their systems, ranging from an R4400-based Indigo2 to an R12K based Power Challenge L. I was usually happy with the hardware, but the sales guys were really slimy, and the company made it very difficult and expensive to get basic OS and compiler updates.

A $3000 Indy might have seemed like a good deal, but when you need a thousand dollars a year worth of hardware and software contracts to support basic administration of the box, it didn't compare too well with its competition.

Of course, my POV is probably severly tainted by the fact that I just did NOT like the sales rep. Half of what came out of his mouth was BS.

On the other hand, this had to have been 10 years ago, and I should probably just get over it.

I thank SGI for a lot of my early career...

[Darth Maul]

It's a shame, because back in the early 90's I got into comp sci and computer graphics, using Indigos and Onyx machines in all my early work. I even bought an Indy for use in college, and there's no way I would have been able to do such cool project work in school without it. It's a shame to see this, because I was as big a fan as SGI could have had back in the day, but I know the day SGI started its decline.

It was SIGGRAPH 2000. New Orleans. I got an invite to the SGI party, and we were all expecting a huge new announcement of a SGI-brand PC graphics card. This would have been the smart move, because about this time PC cards were starting to eat into SGI's markets... So why not use the amazing brand name of SGI and produce a killer PC card? So what did SGI announce? A new line of supercomputers. There were audible groans in the crowd.

Oh well, it was part of history. My Indy still works just fine, and I was even able to update to a newer version of Irix recently... And I'll still wear my SGI shirts, thankyouverymuch ;-).

Re:Sorry to see them go...

[MtViewGuy]

I can cite a few reasons why SGI has fallen on hard times:

1. Current Linux distributions can run workstation level hardware/software.

2. x86-compatible machines now have powerful enough CPU's to run workstation level hardware/software.

3. High-end graphics cards using the nVidia Quadro GPU chipset can do most of what SGI machines can do in terms of graphics but at much lower cost.

Why do you think Dreamworks Animation is using AMD CPU boxes with high-end graphics cards running Linux?

backplane speed?

[YesIAmAScript]

What's the backplane speed you refer to?

This say the GIO64 backplane speed in Indigo2 was 266MB/sec.

This was probably great then, given the limitations of FPM RAM (EDO wasn't even around yet!), but it is peanuts now. Intel's FSBs and AMDs HTs hover at about 30 times this speed now, and there are plenty of slots which exceed this speed too.

Am I missing something? I only looked this up because the amount of time SGI has been out of the loop pretty much means that their systems cannot be anything special compared to current hardware. That doesn't mean they weren't ahead of their time, just that a lot of time has passed and even things that were ahead of their time then are nothing special now.

I had a couple friends who work at SGI and I was heavy into the computer graphics market then. SGI were doomed before they bought Cray. They basically started by taking the work of Evans & Sutherland and bring it to a whole new marketplace. They realized the potential of computer graphics in a broader market, not just defense and similar companies. The problem was, the market was even broader than SGI expected.

Oddly, it was the horrible Matrox Mystique video card that signalled the end for SGI. It wasn't the first 3D PC card, but for many people, it was the first one they owned and used. It ran Tomb Raider with 3d acceleration. These kinds of cards created a whole new market for 3D hardware. This board marketbase pumped money into these companies (Matrox, ATI, S3, and soon after, NVidia) very quickly. And this allowed them to advance their hardware rapidly to the point where a well-equipped PC could match the 3D performance of an SGI box.

SGI was addicted to selling $80K workstations in small numbers, and PCs running 3D Studio Max that could be configured for a bit over $10K just overran them. SGI refused to adapt. Because of their overhead, perhaps it was impossible for SGI to adapt. So SGI was in a marketplace where a 3D workstation could only fetch $10K (and falling), with a business model and overhead (like owning your own CPU designer, writing your own OS) that made it impossible for them to compete.

End of SGI.

I don't understand your assertion that SGI was an internet player. The cost of their systems meant you couldn't afford to buy an SGI for anything that didn't involve heavy graphics, or else you'd be wasting your money. SUN really did rule the roost there, for a while. Until a broad switch to PCs whomped them too.