Silicon Graphics Inc. today issued a melancholy press release confirming that it has been kicked off the New York Stock Exchange, and will now trade as a ‘penny stock’. SGI was the high-bandwidth, visualization-rich, media-savvy computer systems company that flourished in a decade when media pundits clamored for a winner in what they called ‘the convergence space’. SGI obliged, and spared nothing – for a while its budgets were flush for R&D and fancy architecture. “Oh, SGI, we loved you and you screwed up. Bigtime.”
Very sad.
I hate to see companies that had great products go under.
The also made one of the best Unix in that time, Irix
Their website has been the only place I can find a free listing of the c++ STL functionality; it’s not very good, but it beats the half done work everywhere else!
For that alone, they have my brand loyalty
. They really should have jumped onboard with using cheap computers and parallel software for render jobs. I’d bet people would have bought Irix on Intel at a low price for their old junk hardware, and even bought junk hardware from SGI if they’d sold it cheap.
So, now Sun is selling Opteron and SGI is penny stock. Apple is switching to x86… Sadly, it seems Microsoft may be making one of the biggest non-x86 products on the market now: xbox 360. I think I’m gonna cry. (I’m excluding PDA’s and other small devices)
” SGI was the high-bandwidth, visualization-rich, media-savvy computer systems company that flourished in a decade when media pundits clamored for a winner in what they called ‘the convergence space’.”
The problem is that that space isn’t cheap. Parallel processing and cheap hardware only goes so far. I have a cheap GPU in my machine (compared to what SGI uses), and it’s not going to be producing Toy Story anytime soon. Even if one uses SLI.
Um, I believe you use a render farm for final renders and high end gfx cards for real-time previews…
Their website has been the only place I can find a free listing of the c++ STL functionality; it’s not very good, but it beats the half done work everywhere else!
Yea I know, I’ve found their STL docs useful many times too. I guess there’s also http://www.cppreference.com, but it’s not as thorough.
Yea, and see, it’s the less used things and edge things I want to know usually. Stuff like “hey, how do I iterate through a map?” Which, I did guess iterators would work and eventually figured out the syntax, but still… One of these days I should probably buy O’Reilly’s book on the STL. The handbook is useless though, I’ve browsed it…
For C++, I recommend getting a PDF copy of the 14882 standard for $18. Then, get Nicolai Josuttis’ book “The C++ Standard Library”. That will actually tell you how to use the STL.
$18 for a pdf? I think not. Is it like 12,000 pages?
Well, you might as well add “Effective STL” by Scott Meyers. Ofcourse there are good books, and other printed material. I think the keywords were free and online reference.
Worked for sgi a while back. As with all computer companies, there comes a time when you need to re-invent yourself. If you get it right, great. If not, a slow death. The field always seems to know what’s really going on; the number of times I heard them say ‘They’re doing what?’ (eg buying Cray Research, going into NT.) Senior sgi management had no idea about the culture, the philosophy and the real world problems faced by both customer and the field people in trying to deliver a world class product.
For all the mystique that SGI had, their hardware wasn’t that great and the software was so-so. In no time at all high-end PCs surpassed the mighty Silicon Graphics Workstation in sheer performance, and the combination of Windows and numerous 3D graphics software proved to be far more versatile and user friendly than IRIX plus Alias Wavefront.
And in terms of rendering power, SGI simply could not match Linux based render farms, period. Linux may be sucktacular for desktop use, but there is no question it has its place in render farm applications.
So, another mighty competitor fell, crushed by Intel and to some extent Microsoft and Linux as well.
Well, if by “no time at all” you mean in only 15 years. Silicon Graphics was an industry leader for a very long time. Their hardware and software was truly great, but that really was a decade ago, so it is no mystery why they have faded into obscurity in modern times, but claiming that Silicon Graphics wasn’t an important and influential company in the graphics industry for a very long time is just silly.
Of course it seems that the rest of your post implies that you have gotten what Silicon Graphics does a bit wrong, the supercomputing side of things is a newer attempt to change the direction of the company (one of the failed steps). Sure they had a lot of parallel processors quiet some time ago, but that was a matter of managing to run simulations that were to be visualized on the graphics systems. The graphics systems which was Silicon Graphics claim to fame.
It is however really a matter of commoditization, Silicon Graphics could no doubt have continued to push the envelope if there was a need for it, but most visualization work is done quite handily on a PC with a NVidia off-the-shelf graphics card these days. There simply were no problems to scale to beyond a certain point.
The hardware wasn’t that great? Please. Go get the fastest x86 box you can find, and try running multiple streams of uncompressed HDTV off it’s disks to your display in realtime.
Can’t do it? Funny, the 10-year old SGI Octane2 will do it all day long.
x86 CPUs may have advanced past what SGI was able to provide with their MIPS processors, but they havent caught up w/regard to I/O hardware and the bus technology.
Even the lowly O2 still has a place in broadcast graphics for its genlocked OpenGL-to-video capabilities which is notoriously difficult to do without expsive 3rd party broadcast video gear on PC/Macs.
Nvidia etc. wouldn’t be here without SGI. The 3D games/film/video industries as we know them simply wouldn’t exist without SGI and it’s revolutionary hardware.
For many years, and in some areas to this day, no PC, Mac or UNIX workstation could come close to the capabilities of a decent SGI workstation. IRIX on SGI is one of the big reasons why Linux has managed to gain any traction at all in the render-farm space.
If the studios hadn’t had such an IRIX-based workflow, well, guess what – the Renderfarm would be dominated by someone else, most likely Microsoft.
The only reason Linux has traction is because it was capable of acting as a cheap substitute for IRIX.
Personally, I think you are way off base saying the hardware and software is so-so – I can still drag a windows around on my 200Mhz O2 with less flicker visual artifacting than I can on my 3000Mhz Athlon w/Nividia card. It captures and plays back video more stably than any Windows PC i’ve used, and it’s X server is capable of the same types of whizz-bang OpenGL integrated effects as OS X is.
Unfortunately, SGI’s dysfunctional management just failed to focus their R&D efforts where it mattered, eneded up alienating most of their talent and fell into the trap of trying to compete directly with commodity x86 manufacturers.
The CEO that killed SGI (Rick Belluzo) then jumped ship and went to work at Microsoft immediately after squandering all SGIs resources on a line of workstations that were reviled by everyone except PC World etc. reviewers who thought they were neat simply because they ran Windows NT.
It’s sad that SGI is no longer a major player, but I dont see that their demise had anything whatsoever to do with the quality of their own hardware or software.
We may have PCs with thousand dollar SGI-inspired video cards, Macs that make a centerpiece of SGI OpenGL effects, and Linux boxes running rendering software pioneered by SGI but we’ll never see innovation like SGI gave us from Intel, AMD or Dell, and the idea that a company like Microsoft could build a piece of software like Maya from the ground up is laughable.
What we have now isn’t a step forward, and the computer world is not a better place without SGI.
try running multiple streams of uncompressed HDTV off it’s disks to your display in realtime.
Can’t do it? Funny, the 10-year old SGI Octane2 will do it all day long.
And this is why SGI owned that market 10 years ago. They were the only company making computers with that kind of insane internal bandwidth. And you PAID for it. (Hoo BOY did you pay for it!)
However, in the past 5-6 years, new technologies and designs have allowed off the shelf PCs to catch up or come close enough.
Even the lowly O2 still has a place in broadcast graphics for its genlocked OpenGL-to-video capabilities which is notoriously difficult to do without expsive 3rd party broadcast video gear on PC/Macs.
Now, you work at a TV station and the O2 has died a horrible death.
A Quad PowerMac + gear will be less than a comperable replacement from SGI.
Yes, I think it’s a damn shame that SGI is hemoraging money. They have built some of the most sexy computers (both in form and function) known to man, and I’d love to have a Tezro, just ’cause I think the box is so damn pretty. (In 10 years I’ll be trawling eBay for one …)
But damn if SGI didn’t keep shooting themselves in the foot over and over and over.
(And hell, if I were driving the bus at SGI, I’d slash the price on the desktop/workstations to bring them more in line with a top end Apple.)
SGI hardware was all top end. Powersupplies, Motherboards, 3d Graphics. All extrodinary. The FREAKING CASES ARE ROCK SOLID!
Q: What machine can hold the largest amount of ram?
A: SGI Origin 2000. 2 TB of ram, which can be partitioned amoung up to 512 processors. NASA Goddard still has one of these, because the Linux Clusters still have not caught up for atmospheric research. All closely coupled CcNuma arcitecture.
My Indigo Dual is still in reach. It still can render, play a Video stream and rotate complex objects without a hichup. ( You can see the specs on Dusty Computing )
I have never seen a PC do anything without a hickup, with the exception of a BeBOX.
SGI’s stock value dropped -31.82% in value today. From $0.66 to $0.45.
The stock soared a bit earlier this year, which is why I was following it, but the rules are that you have to stay above $1 average value for a 30 day consecutive period.
http://www.osnews.com/comment.php?news_id=6306&offset=30&rows=45
…I certainly met some fierce opposition at the time, but yes, this was clearly inevitable, that much was evident well over a year ago (that story was from March 2004)
The company still had 800 million in revenue this year. Yet they had a 100million loss. You would think the execs would see big red flashing lights when they are losing 100 million dollars, and then decide to do major cutbacks
I have 100 shares of SGI that i bought at $1.26 hopefully someone buys them.
They failed because they went Linux…
They failed because they thought that Linux was the way to go…
They failed because they thought Linux was stable enough for them to use…
They failed because they thought people and investors would fall for the Linux and F/OSS buzzwords…
They are just one of the companies that have learned the lesson that Linux is not the solution to their problems, with many more companies to follow.
You mean… IBM is the next?! OMGWTFBBQ
P.S.: you’d better check the history of SGI. The linux transition came too late, and after being backstabbed by MS on an agreement to share technology and release SGI Windows NT workstations.
MS stabbing someone in the back?! That never happens. I hope they find the leadership they need to pull themselves out of this immense hole their in.
For some reason I feel gutted about this. They were always the shiznit in graphics…
Linux transition came too late?
More like it came way too early for it to be a viable path to go at all.
Blame other companies all you want, but SGI’s demise is their own doing.
Please, learn to read darling; he pointed out that converting to Linux to solve financial problems isn’t the solution, the panacea that will deliver rocketing profits overnight; if you’re crap without Linux, you’ll still be crap after adopting and basing the whole business model around it.
The fact remains, SGI went for the wrong team; they went Itanium but had no well known, high profile operating system to bolt ontop of it – Linux STILL has things lacking that are available in commercial UNIX’s.
SGI would have been better off, for example, working with SUN to port Solaris to Itanium or POWER and concerntrating on their bread and butter – but personally, I think the biggest mistake was the move to Itanium; POWER would have been a better choice, coupld that with NUMAFlex and Solaris, and you would have all the ingredients for a great machine.
They also failed to see the writing on the wall, NO ONE, I repeat, NO ONE is going to pay $14,000 for a workstation that performs worse than a run of the mill x86 workstation and worse, lacks software that is supported by the vendor to a decent level.
Add that the fact that SGI had no big players on their ISV books – hell, its been atlest 2 decades since Adobe released something for Irix, little wonder the niche has been getting smaller; there are only a small amount of niche markets before they’re flooded – sorry, its either that they join the mainstream, with a dabble in the niche markets; no company can survive simply going for small fry; SUN has learned that, and are FINALLY getting out there and realising that EVERYONE is their customer, not just a handful of loyal clients.
s/Linux/Itanium/
They failed because they went Linux…
Wrong little MS astroturfing buddy. They failed because they went with Windows NT in the mid 90’s after apppointing a MS backed saboteur as CEO, who had just left HP after nearly destroying UNIX there on behalf of his masters in Redmond. He went on to become a V.P. at MS but he was of little use to them there so they dispensed with him like a pair old dirty socks.
Linux was SGI’s only hope after that mess
Linux could neither help nor harm SGI, since its problems had nothing to do with using Irix, NT, or Linux. SGI like the other major UNIX workstation and server vendors, saw an artificial demand during the bubble and a subsequent rapid erosion as the bubble waned. When purchases were made the costs associated with the niche hardware were unpopular with the new-found frugality of technology companies, and “good-enough” x86 systems emerged on the workstation market with superior compute-performance and significantly cheaper nVidia video cards with firmware optimized less for gaming and more for CAD, and in the server market with price/performance savings. It doesn’t matter if large margins are obtained from hardware strategies with technical advantages if those advantages are only necessary for a niche whose volume is insufficient for meeting the overhead associated with operating a company that has scaled its operations under the assumption that the dot com era would never end. SGI, like most other UNIX vendors did this, and suffered tremendously as a result. Attempts to recover have met with less than stunning success.
Please, a shot in the back of the (empty) head for Marcellus..
I once worked for a company and our team designed & manufactured power converters for the SGI UNIX Workstations. The specs SGI wanted had such incredibly tight tolerances, those of us on the design team were shaking our heads. Not many of us understood why they needed to be so tight. One of the SGI engineers dropped off a couple of workstations for us to poke into to marry up our converters.
After tearing into them, the SGI Unix workstations quickly earned the term, ‘battletanks,’ by all of us in the team because they were so over-built and designed so well. Once we started dissecting the circuits our power converters would connect to, we quickly learned the ‘why’ regarding the tight specs.
A couple of years later I noted that the movie, “Jurassic Park,” had SGI workstations that controlled the, ‘park.’ I wondered if the ones in the movie had any of the converters we built. Probably not, but it was good to see them in the movie.
Sorry to see them tajken off the NYSE.
1) Failure to develop IRIX further. Folks, this was once the *premire* 64bit *nix based OS. It had a beautiful, easy to use GUI (compared to anything else out there.)
2) Failure to stay abrest of market trends regarding hardware price/performance. For years, if you wanted 64 bit computing, you had 2 “big” names: Sun and SGI. And hoo-boy did you pay for that ability to do heavy lifting.
Now there’s AMD and IBM’s PPC 64bit solutions — Cheap cheap cheap cheap compared to what it costs to walk in the door at SGI (or Sun)
Why should I pay more than my car for an SGI machine when I can get the same, or damn near the same, at 1/3 price?
3) Failure to stay abrest of trends regarding Software.
3a)They let IRIX stagnate, and ignored Linux too long. They should’ve taken Linux and made a great IRIX replacement … 5 years ago. They could *own* the 64 bit Linux market with a mature, polished, distro at this point. But no.
3b)Failure to see what various distributed computing software solutions could do to their business. It’s now cheap and fairly easy to cobble together a render farm/super computer.
4) Failure to stay abrest of market trends, period.
SGI *still* prices its hardware/software as if it’s the big fish in the pond for high performance 64 bit graphics computing and these are the late 1980’s/early-mid 1990s of superfat corporate budgets.
Following the dot.com crash and the recession, budgets have been slashed, and if damn near the same thing can be gotten for 1/2 to 1/3 of the price, that’s what gets bought.
—
I mean, really, people bitch about Apple’s PowerMac prices, but within the next 4 months I plan to get a Power Mac G5, and after I’m done adding an extra drive mount and PCI-E SATA contoller card, I’m going to have a machine that can run neck and neck with any of SGI’s desktop workstations, and runs a lot more software — at half the price.
kadymae hit it straight on the head.
The people placing all the blame on Linux or NT exclusively know very little about the company and its history.
There were a laundry list of reasons for SGI failing but that is the way with most companies.
The word “beleagured” comes to mind.
2 years ago i was on a contract job in a military base
they had just hired a person to build pcs for them, these pcs were part of a migration
they were migrating from sgi machines to linux on regular pcs with nvidia cards
they said that linux on a pc was more than capable to do what the sgi machines could do at a fraction of the cost
they had stacks of sgi machines sitting up against the walls, slated to be destroyed
anyways, i thought that was interesting….
And I just completed a transition from 10 Sgi Workstations to dual Opteron IBM Intellistation Workstations paid entirely from the 3-year service budget of the Sgis. In other words, we get faster upgradable machines, good graphics (Quadro FX 3000), for zero additional cost.
The trend is inevitable.
ILM did the same thing as I recall. They switched to Linux machines running NVIDIA graphics cards, which were cheaper and faster than the SGIs they replaced.
Time to buy out SGI and get all the video architecture capabilities and throw it into the high end macs 😉
While you’re at it – open source IRIX so it doesn’t go to waste :p
SGI isn’t dead YET! Not too long ago they went for ‘powercomputing for the masses’, remember?
well, its not *the* end. But close…
You will be missed, SGI! *sniff* ;'(
I used to have to admin IRIX in a shop that had IRIX, HP/UX, Solaris, and Linux. IRIX was OK, but generally behind with respect to standard UNIX stuff: DNS, NIS, NFS, Sendmail, etc.. Sometimes it was hard to get IRIX to play nice with the rest of the network.
We used O2 machines for doing PCB development. They were cool and all, but in the end not worth the money. We got rid of the last one in 2000. We were able to replace them with Windows machines running the same software for 1/4 the price.
Well, this doesn’t mean they cannot come back and be as they were before or even better; think about it they just were dropped from stock market but their company is not bankrupted. They could do major reform, hire new CEO, use linux and jump start again. I think apple had rough times like them before but they managed to survive after all.
I will be unhappy to see an excellent player fall down.
Linux isn’t a pancea for ailing companies. It wasn’t for Apple.
I think apple had rough times like them before but they managed to survive after all.
Apple was never in this much trouble. SGI’s market cap right now is about $200 million and they have almost no cash reserves; when Apple was in deep, deep trouble they had a market cap running between $6-10 billion with massive cash reserves. Once you’ve been delisted from the stock exchange, it is very difficult to get back on your feet. I hope that SGI manages to pull through. I own an Indy and it’s a very cool machine. If this is the end for SGI, hopefully the engineers can find a home at Apple.
And we still have it doing great cars (like the PT Cruiser).
Maybe SGI needs a hero, too — another Iacocca.
Good luck to them. People who make quality products will always have a warm place in my heart.
dô-ôb
Oh dear, the PT Cruiser is a peice of junk. Much better to hold up their 300M and other recent vehicles. The PT Cruiser/Prowler were pretty attempts but woefully underpowered and had quite a few service problems.
…Apple should buy them
Now there’s AMD and IBM’s PPC 64bit solutions — Cheap cheap cheap cheap compared to what it costs to walk in the door at SGI (or Sun)
Can you back the IBM PPC solution with actual prices or is this another one of the IBM fanboys PPC blowjobs?
Get thee to the IBM website and check the prices.
And you can get to them with less digging than on the SGI site.
(And I’m a fangirl thank you very much.)
Anonymous Fud Guy asked:
Can you back the IBM PPC solution with actual prices or is this another one of the IBM fanboys PPC blowjobs?
Blades Start under $2,500
http://www-132.ibm.com/webapp/wcs/stores/servlet/CategoryDisplay?ca…
Open Power starts under $3,500
http://www-03.ibm.com/servers/eserver/openpower/hardware/710_browse…
Power 5 Running AIX starts under $5000
http://www-03.ibm.com/servers/eserver/pseries/hardware/entry/
At SGI, the low end starts just under $7000, but you have to call for exact pricing and specification details
http://www.sgi.com/products/servers/altix/330/
Then there are those workstations from Apple, coupled with MacOS X and the mountain of applications that are available.
Crying shame for a really great company that made some really great stuff happen.
…SGI has been a penny stock.
http://finance.yahoo.com/q/bc?s=SGI&t=5y
Anonymous Fud Guy replies:
Blades Start under $2,500
plus blade chassis which starts at 1,999
http://www-132.ibm.com/webapp/wcs/stores/servlet/CategoryDisplay?ca…
Open Power starts under $3,500
which you need to request a quote for. Sounds like SGI. The equivalent Sun box (which you neglected to mention) starts at $3200
http://www.sun.com/servers/entry/v210/ with quad gigabit ethernet rather than the dual in the power.
Power 5 Running AIX starts under $5000
and repeating myself
which you need to request a quote for. Sounds like SGI. The equivalent Sun box (which you neglected to mention) starts at $3200
http://www.sun.com/servers/entry/v210/ with quad gigabit ethernet rather than the dual in the power.
Now my point is not to harp on about SGI or Sun machines, my point is that you are saying power is cheaper for everything and it is not. IBM are pushing Linux on power for hardware sales, not for the love of Linux. I just wish the IBM fan boys (and girls) could see this.
– Anonymous Fud guy who has lost a previous job due to IBM professional services (which is why I am anonymous)
“IBM are pushing Linux on power for hardware sales, not for the love of Linux.”
No shit? HP is pushing Linux because they are good company? OH wait…
Come on, a v210 with a 1.3ghz UltraIIIi is comparable to a 1.5ghz Power5? By most benchmarks I’ve seen, the Power5 performs twice as fast as a comparable speed UltraIV. The IIIi isn’t even in the same ballpark. We’re talking similar price from IBM for 2-3x the performance.
I think Apple should buy them and team up with NVidia, IBM or HP to deploy some nasty Power or Itanium graphics workstations.
In these recession times, when fast networks and x86 processors provide cheap and scalable distributed computing, if there is any market for high-end RISC workstations then it’s an entertainment industry. (Though neither Power or Itanium are pure RISC architectures in their concepts, like MIPS was.)
For good or bad, IRIX is certainly dead as an operating system. Near future of visualisation and FX is probably in OSX while GNU/Linux is yet to be established in this field.
Just my thoughts.
I am surprised SGI made it this long, they haven’t been profitable for years. Old military contracts have been keeping SGI on life support.
See a trend here? Consider UNIX pure plays:
– Sunw is not doing well. Recently the board voted to do away with the poison pill. In other words, they now hope somebody will buy sunw.
– SGI being delisted
– Scox out of business and bankrupt within two years – only msft fud money has kept it on life-support this long.
Then consider:
– IBM seems to be moving away from AIX favoring Linux.
– HP hasn’t done diddly-squat for HP/UX is the last five years.
UNIX to become VMS?
Lessons to be learned here, I think. UNIX companies needed to get it together ten years ago. UNIX would have totally stomped NT if they had. Let’s hope Linux
doesn’t make the same mistakes.
Um no……Steve learned his lesson with NeXT.
Hardware is a commodity, it’s all about the software.
Especially more so now that Apple is moving towards more ‘off the shelf’ components in Intel.
The last thing they would do is aquire a highly proprietary hardware platform.
Not that I agree with Steve mind you – NeXT’s were killer boxes, and Intel’s processors are a joke compared to AMD or IBM.
Most of the commercial UNIX vendors are either in deep trouble, or phasing out their UNIX line of products, but OS X keeps marching on and growing bigger. (Technically OS X isn’t UNIX, but that has to do with copyright, not technology. Underneath Aqua, OS X is UNIX) UNIX isn’t dead yet. Hell, Solaris might even make a comeback if Sun amangement pulled their heads out of their asses.