SGI issued its most ominous regulatory filing to date, warning that a bad 2006 could force the former high-flyer into bankruptcy. In order to improve its business, SGI will consider measures ranging from axing or selling off product lines to pursuing ‘a strategic partner or acquirer.’ The hardware maker will basically look at anything and everything to remain a going concern.
Geez, what happened to this company? They were on the top of the multimedia UNIX workstation world with the Indy workstations and they appear to have lost it all.
they bet the company on linux on itanium and threw everything that made them unique out of the window.
That is exactly what happened, and no matter how much hype Linux get, I can’t understand people in decision making roles start reading between the lines and analyze what Linux means for their business.
for SGI it obviously meant a definite end (as their previous product lines were obviously unique etc).
for Novell I’m confident we’ll see their business loose in the longer run.
Palm is going the exact same route any minute now. The same second they announced “We’ll skip our unique approach with PalmOS and hit the general trend with Windows and Linux” the company has lost any value it previously had.
Sun has tried the same with poor results…
Who’s next???
When will business managers realize that differentiation is actually something good which builds value….sheeeeshh…
for Novell I’m confident we’ll see their business loose in the longer run.
novel does already make red numbers
for sgi they should have kept irix/mips for contentcreation and other high-level tasks. for those low workingmachines in renderfarms bsd or linux on x64 would have been a good decision.
finaly sgi learned the hard way that enhencing linux is helping your competitors.
i still hope that it’s not too late for them…
Edited 2006-02-09 22:47
What happened was they underestimated Intel and Nvidia. With a P3 600 and a Geforce card running Maya you all of a sudden had a workstation that was good enough for most of your artists needs at 1/10 of the price. This totaly gutted their workstaion market. MIPS could never catch up and for every new CPU Intel came out and every new card from Nvidia there where less and less places where sgi had an advantage, yet they where still charging 10 times the price. Argue quality and stability all you want, people simply didn’t want to pay.
On the server side clusters started being able to do more and more things which before needed a single large server. Sure clusters took up more space, drew more power and where harder to set up and administer, but the upp front price difference was so huge that it was still worthwhile.
Basically they had to get off MIPS, since MIPS couldn’t keep up with Intel. They decided that porting the necessary Irix bits to Linux was easier than porting Irix to Itanium. Unfortunatly Itanium never lived up to its promise.
they bet the company on linux on itanium and threw everything that made them unique out of the window.
No – Before that they had a CEO who, while working for HP, tried to destroy Unix there. When he moved to SGI he had them move to Windows Workstations and Intel kit (Ugh!). After this nearly destroyed the company, together with him selling most of SGI’s patents to MS, he left to receive his payoff and was appointed a VP at MS. They then decided he was no longer of use to MS and gave him the boot.
The Linux strategy was the companies only hope to escape from this sorry sad story of treachery and it was probably too late.
Edited 2006-02-10 13:52
IMO it would be great if Apple bought SGI and integrated its technology into its products. Or even keep it as a spinoff, separately run but complimentary. This is terrible news though, SGI was once one of my dream companies. It’s a pity it’s come to this.
Does SGI have anything left that people want?
Does SGI have anything left that people want?
If you want a computer with 512 CPUs and 128TB RAM, then yes
Their visualization systems are still great, but the market isn’t very big.
EmmEff: Does SGI have anything left that people want?
stew: If you want a computer with 512 CPUs and 128TB RAM, then yes
Their visualization systems are still great, but the market isn’t very big.
whilst that 512 processors sounds interesting, the reality is that 512 processors & 128TB ram in one machine is precisely what ‘people’ don’t want for 99.999%
‘people’ in that market want flexibilty, they want clusters because clusters are better performant and desperately cheaper to purchase and maintain
>Does SGI have anything left that people want?
Patents on 3D technology. I would love to see Microsoft buy all the Patents out as they would own the entire 3D industry and this website would go nuts.
It would be just interesting to see the reaction. SGI own’s the entire industry with their patents. Remember, they were first out around 1980-81 so they would have a lot of technology patents that nobody else has.
The first sign of real problems was when the Playstation came out and PC hardware started coming out with GPU’s in 1998.
You have *got* to be kidding, right?
“It would be just interesting to see the reaction. SGI own’s the entire industry with their patents” – well, let’s see :
a) They can’t obviously own the industry as they are losing probably 90% of the deals in the 3D space (Asia being the exception)
b) Their patents can be worth much, as SGI otherwise (in the light of the above) would have defended the patents
c) The current and by far most interesting 3D patents are today not held by SGI, but by Nvidia/3DLabs/ATI in general and companies such as Siemens/TeraRecon etc in medical imaging, Halliburton/Schlumberger etc in oil and gas and so on.
Remember, SGI is debt-ridden beyond recognition. Their debt is likely worth much more than whatever assets they might have. A lot of their IP has already left and walked out the door. The only ones who have a strong interest to keep SGI alive is US defence and Government. Most others can do very well without SGI, thank you.
Patents anyone?
This is sad news to me. I always wished they pushed Irix as a Linux front end.
Wasn’t IRIX unofficially killed of a year or so ago?
Too bad for SGI, they had some pretty good stuff. I am left wondering who their remaining customers are.
If a company buys them out (Sun or Apple since these two have not “Diversified” as much as IBM) then the company that buys them out can hold on that that niche that SGI has and build up on it.
I think that Apple is better suited to aquire them, given their whole Pro range of Apps, but Sun could be a good contender as well as a means to diversify their “the network is the computer” business
Hopefully, If someone does gobble up SGI or they do end up going out of business (bankruptcy does not necessarily mean going out of business). IRIX will atleast get open sourced, and not let it just fizzle out and die. I’m sure there is plenty of good tech in IRIX that the other remaining UNIX vendors haven’t thought of or implimented yet (or ever). I’m sure they would be plenty of people that would love to hack away at IRIX. Here’s to hoping.
Irix is still supported, it’s just not #1 anymore. They’ll keep pushing minor updates for a while. 6.5.29 should be out soon, but don’t expect an Irix 6.6 or 7.0 ever.
Someone started making a 4dwm like front end for Linux called 5dwm:
http://www.5dwm.org. Shame, it seems dead.
We has sold a lot of SGI servers and they are still the most stabile servers i have ever worked with (sun servers on second place). I really loved irix and it´s little “toolchest”, so clena and fast and stabile.
Also, thier XFS filesystem, one of the fastest filesystems i have worked with.
This is sad to hear. I thought their business with governments and military was good business…
Maybe SGI should buy Palm. Then the BeOS could continue its tradition of being owned by companies with useless management. They could create the ultimate media appliance, running multiple instances of BeOS on a virtual machine running on IRIX powered by Itanium ‘for ultimate speed and stability’. Only $6000, at a computer store near you (monitor and tv tuner not included)!
When you spin off MIPS, sell Alias/Wavefront, drop IRIX and then embrace Titanium back in the days when no one would touch it then that’s a tall order to recover from and prosper.
“sell Alias/Wavefront”
Now, _that_ was a mistake…
Had they had any this would not have happened. They might want to see if they could borrow Steve Jobs for awhile.
Steve Jobs is great at running a desktop computer company like apple. but trying to run a company like sgi might not go all that well since apple’s server market never really picked up.
Noooooo SGI. Bye-bye-bye Intel kisser !!!
For those who don’t know… many of the “supercomputing” features in the Linux kernel are from SGI software engineers. I don’t know, small features like Non Uniform Memory Access (NUMA) just to drop names. SGI has helped linux scale up to many many processors. If nothing else, maybe IBM will swallow SGI and keep some of their developers.
for betting the farm on Linux and Itanium. Exactly what “secret sauce” did they have left?
Apple and Microsoft are laughing their a**es all the way to the bank when they see Linux become a pac-man and slowly eat at Sun, Novell and HP.
Edited 2006-02-09 22:37
Apple and Microsoft are laughing their a**es all the way to the bank when they see Linux become a pac-man and slowly eat at Sun, Novell and HP.
How so? Solaris is now price competitive with Linux and in terms of features, Linux can’t even hold a candle to it – which I ask as a side note, the whole point of Linux was its UNIX like qualities, its relative stability and ability to run on generic x86 hardware at a low cost; now that Solaris provides all that, apart from maybe the desktop (the real focus for Linus – ask him yourself if you don’t believe me), then what is the purpose of Linux?
Novell on the other hand, isn’t as far up shit creek as the media would like to make out, unlike SGI, atleast they didn’t go in multiple directions, they went straight from Novell to SuSE Linux – a clean transfer vs the swing from NT to Linux to Irix, then back again to Linux that SGI did, all whilst their product line died in the ass, customers left to IBM and SUN for their scientific and engineering needs.
Sad to say this, but SGI has gone past the point of no return – they had a chance to turn things around, had they not wasted their time on Windows NT and the whole x86 debarcle, kept their assets such as MIPS etc. SGI right now would be, admitingly not exactly raking in cash, but they atleast would have been able to keep their head above water.
Sorry, selling assets and cutting staff has NEVER been a successful strategy for making a business profitable for the long term – to turn around the business you need revenue and profit GROWTH.
One of the main reasons that SGI moved from MIPS to Itanium was to lower expenses. Intel promised a lot in Itanium and SGI was taken by the stories from both Intel and HP.
One could safely say that SGI just happened to be unlucky and got screwed for believing the hype.
The sad thing is that, what happens to OpenGL now ?
One of the main reasons that SGI moved from MIPS to Itanium was to lower expenses. Intel promised a lot in Itanium and SGI was taken by the stories from both Intel and HP.
Yes, but they sold their MIPS processor business – they developed and MADE processors – at the very least, like I said in a previous forum reply, had everything gone tits up in the Itanium venture, they would have atleast had their MIPS embedded processor business along with Alias Wave Front to fall back on.
The move to Itanium wasn’t a bad move persay, what was a bad move was selling all the crown jewels off before Itanium prove/disproved the hype.
One could safely say that SGI just happened to be unlucky and got screwed for believing the hype.
Not really, there have already been proven examples in the past of peopl trying to implement VLIW ISA processors and failed – Intel is one of them with their i860, MAJC developed by SUN is another one. So there is a train wreckage that SGI could have learned from.
The sad thing is that, what happens to OpenGL now ?
Nothing, because most of the patents are now held by Microsoft and the committee that overseas its development is so enept that the net result has been DirectX over taking OpenGL meaning anything thing that isn’t Windows based is screwed when it comes to portability of code.
The sad thing is that, what happens to OpenGL now ?
Nothing, because most of the patents are now held by Microsoft and the committee that overseas its development is so enept that the net result has been DirectX over taking OpenGL meaning anything thing that isn’t Windows based is screwed when it comes to portability of code.
OpenGL is a real concern. DirectX (Actually, Direct3D/DirectDraw which is just a part of the full DirectX API) is good at games and might be better or more featureful than OGL 1.1 but when it comes to virtualization and/or 3D modelling suites, which relies heavily on the OpenGL API, the newest OpenGL is king. I´ve seen some graphic cards sold to real professionals that don´t even have DirectX support on them, for starters.
DirectX is a no go on Macs and other unixes that share a respectable slice on the workstation market. Its a well known fact that a few Dreamworks and Pixar´s flicks have been made on Linux workstations… That´s right! Not Linux render farms, I said artists workstations. I can´t quite recall if it was Disney that went extra lenghts to ensure that Photoshop could be run on top of Wine so that their artists could use it on Linux, but I´m sure that one of the big studios did it. Pixar did a test drive for a while but, being Steve Jobs one of their head execs, they started to use Macs mainly over there.
All these studios did that because Unix is good at crouching numbers (making a good renderfarm, of course) but also because it increased their productivity. Tasks that could take several seconds to be completed on Windows were being completed almost on real-time on Linux. No vendor that´s worth their salt would risk not being able to sell to this hugely profitable niche (Please remember that we´re talking about graphics workstations here!). Not to mention that they can´t just throw away all the time, money and effort spent in order to build their software on top of OpenGL.
Edited 2006-02-10 06:43
Every time an SGI article pops out, there come the whiners, blabbing how SGI screwed itself by dopping MIPS/Irix and going to Linux/Itanium…
SGI screwed itself when it didn’t realize that the x86 camp was eating their cake. SGI was used to selling expensive machines to the graphics folks, and then those folks found out that a pile of cheap boxes with Linux could do the same work as a pile of expensive (albeit good looking) boxes from SGI, coupled with a pile of licenses for IRIX.
SGI should have diversified their offer. They could have pushed x86 servers and workstations with the SGI brand to keep HP at bay.
They did try that Visual Workstation stunt, but that had a couple of problems: they had an x86 inside, but they were different enough to not be a PC and have big price tag. The idea was there, and it was good, but the implementation sucked.
But the real mistakes were the selling of MIPS and Alias.
MIPS had no future in servers or workstations, but it is very popular in the embedded space. They could have been designing embedded processors, and collecting royalties from every one of the millions of appliances out there (just imagine how many Linksys routers there are).
Alias, well, that seems to keep being profitable.
And they could keep selling supercomputers… But since that’s their only business right now, if it fails, they’re dead.
SGI should have diversified their offer. They could have pushed x86 servers and workstations with the SGI brand to keep HP at bay.
They did try that Visual Workstation stunt, but that had a couple of problems: they had an x86 inside, but they were different enough to not be a PC and have big price tag. The idea was there, and it was good, but the implementation sucked.
Yes, that might have worked. After all, Apple and Sun are doing the same: Taking off the shelf components, putting them together in a design without out 8088 backwards compatibility, put them in a pretty case and install a decent Unix. Imagine a dual Opteron with a Quadro FX card in a Tezro case running Irix
SGI mainly failed before going to Itanium, it would have made no difference where they went.
The real world of 3D has already moved on. You can get cheaper servers just by going with 80×86 and a GPU and even network them.
By 2007, we will have PC’s with 64-bit processors with four cores and a Direct X 10 based GPU that also does four SLI cards all on PCI EXpress 2.0 busses.
Then you can network this using a gigabit ethernet to have a render farm.
The prices of all that will be lower than what SGI could ever have done.
Oh no there is never Linux bashing on OSNews! Never! Yeah right. That’s all I hear and then I see this BS. So exactly how did Linux kill SGI again? Did IRIX power their super computers? Was the Linux operating system responsible for changing the direction of the company? I don’t think so. This has nothing to do with Linux. SGI decided to totally change directions away from graphic workstations to super computers. If they stayed with IRIX nothing would have changed. They would have ended up with a UNIX that couldn’t run on super computers and a UNIX that was overpriced. They didn’t have anything left to offer.
Which begs to question; why did SGI sell off Cray? it seems Cray is doing a better job at making money off AMD64’s that SGI ever do with their MIPS/Itanium machines.
As for the Linux scapegoat, its commonly used by company fanboys who try to look for excuses to management incompentcy – like you said, SGI was up shit creek long ago, and sticking with MIPS would have been brain dead HOWEVER like I keep saying, had they kept all the assets they sold off, waited it out and then moved to the AMD64 platform, it would be a different situation entirely.
As for the operating system to use, Solaris would be a good one that comes to mind – large network of ISV’s, opensource (recently) and very organised plus it has the whole ‘real UNIX’ aura that comes with it.
But as I said previously, they’re up shit creek and there isn’t a thing they can do but pray and hope someone will buy them out.
Typical. I speak the truth and get modded down. No one even had the guts to respond to my claims because there is no truth in the claim that switching to Linux killed SGI. In fact Linux would have killed SGI from the outside if they didn’t switch, and not because of licensing or hype or any of that. Linux would have killed SGI because of capitalism. In any business if someone comes along and can do the same thing for cheaper then you’re in trouble and you need to react, the same way that Intel has made inroads in the server market. Are x86 processors superior in technology? No, but they can do the same thing for cheaper. Linux + x86 has taken a lot of server market from big iron and will continue to do so. As long as that happens I’m sure we’ll hear all about the “viral” GPL and how it is equivalent to communism and such, while reality flies over the heads of critics at 100mph.
Victum from Linux?!
>OpenGL is a real concern. DirectX (Actually,
>Direct3D/DirectDraw which is just a part of the full >DirectX API) is good at games and might be better or >more featureful than OGL 1.1 but when it comes to
>virtualization and/or 3D modelling suites, which
>relies heavily on the OpenGL API, the newest OpenGL
>is king. I´ve seen some graphic cards sold to real
>professionals that don´t even have DirectX support
>on them, for starters.
With Vista supporting 64-bit, multi-core processors, Direct X 10, Brand new driver support, PCI-EXpress bus 2.0 and better security and stability who needs OpenGL 2.0?
o Direct X 10 has been re-written from the ground up
for speed and very low overhead for the latest programmable technology. No more cap bits, so if hardware is a DX 10 video card it has to have every feature promised by DX 10 and has to pass tests to make it so. DX 10 is also more programable and is more flexible and offers a lot of technology that OGL 2.0 does not. Also DX 10 is just as open ended as OGL.
o Windows Vista drivers have also been re-written from the ground up and contain no legacy code or devices, therefore are much faster now and are in usermode now and not in the kernel. Which makes things 1000 percent more stable and secure.
o 64-bit CPU’s with multi-core are supported now so you could have a processor that has four cores that is a 64-bit CPU with a 64-bit OS, memory and speed along with plenty of registers should no longer be a problem. DX 10 also works with these features as do the new drivers for DX 10.
o New windows 2-D/3-D animation (Windows Presentation Foundation – i.e. Flash killer) uses DX and can use DX 10 for even more effects. This will be used in and built-in future versions of IE (Internet Explorer 8 probably). This is built right into vista and uses an XML type language. So future versions of windows apps will use this as well.
o Maya has been bought by the the same company that puts out 3D Max which is mainly a Windows 3D application company.
o PCI Express Bus 2.0 supports 10 Gigabytes per second and vista is designed around this bus along with DX 10.
OpenGL will work with consoles are closed devices, but they are mainly dead when talking about computers (beyond the Macintosh at least).
When Carmack is now using DX, you know OGL use is going to be fairly limited.
By 2007 OpenGL will be done and you can put a stick in it. You will still see the open source comunity work with it, but for other things, it’s days are numbered.
OpenGL is dead when it comes to games, but visualization and high end content creation is still, largely, a unix market, and even the apps that run on windows in that space also run on *nix. It’s most likely going to take longer than 2007 for MS to gain that market.
>OpenGL is dead when it comes to games, but
>visualization and high end content creation is
>still, largely, a unix market, and even the apps
>that run on windows in that space also run on *nix.
>It’s most likely going to take longer than 2007 for
>MS to gain that market.
Well, I don’t know about this. With faster results, better driver support, better feature support and feature support that works with any DX 10 video card, 64-bit driver support, more support for newer up and coming formats I would think that windows will take over for that market as well.
I feel like it’s just a matter of time. Windows Vista in theory should have a lower TCO than older versions of windows and thus is more stable and secure.
Also the fact that AutoDesk now owns Maya and AutoDesk is mainly a Windows developer with 3DS MAX and they bought out Maya.
Let’s hope Sun picks whatever will be left of SGI, at least Sun is the only company out there that will be able to outsource IRIX after assuming the ownership. SGI dosn’t have a whole lot to offer to Sun, but who knows if the price is right before they go bankrup, Sun might pick them up…
> What happened was they underestimated Intel and Nvidia. With a P3 600 and a Geforce card running Maya you all of a sudden had a workstation that was good enough for most of your artists needs at 1/10 of the price. This totaly gutted their workstaion market. MIPS could never catch up
This is a load of crap, SGI could more than hold itself with MIPS and their line of workstations against Intel in the high end 3D and postproduction markets. They didn’t have to make their workstations the fastest of the bunch, the workstations just had to be “fast enough” for the job. Plus processors were never the forte of SGI workstationss, it is the accelerated video hardware that was the real value add. If SGI kept on developing MIPS and stuck to their guns, Hollywood would still be powered by SGI and you wouldn’t find a single TV/Film outfit that wouldn’t run SGI as in the good old days. I can tell because a 5 year old SGI O2 is still a much better and more capable video editing platform than any of the most modern PC hardware — hell, video playback for high-res is still unbeantable on even the dustiest of O2’s and Octanes. SGI wimped out and pulled the plug on their own products and spelled doom for themselves by following Intel and Linux. If they fought and did not bend over for Intel and MS, they would most definitely still be in the game — may be they wouldn’t be flourishing but thy would have held their own.
Yes SGI could hold its own, but when you’re charging 10 times what your competitor is charging you had better do a lot more than simply hold your own. I mean for the price of a CPU upgrade to an Octane you could get a new high end intel workstation
I worked for a 3D and post house and ever since the P3’s and Geforce/Quadro showed up, more and more of the work we did could be done just fine intel machines costing a fraction. Yes the O2’s and Octanes where damn nice machines and did what they did best better than anyone, but they where too expensive for too little performance gain.
Sgi rested on their laurels and every year that passed the niche where sgi was the best grew smaller and smaller. And even in the areas where they where the best more and more people where willing to settle for second best since the best was far too expensive.
Could they have fought back with MIPS/Irix on the workstation front? It’s impossible to say, but personally I doubt it. It was simply too expensive and the rest of the tech world was improving too fast.
Webforce is down!! Over…
I was reading how many of their people left to join firms like ATI. That got me thinking (and you probably can guess where I was going with this) that perhaps SGI missed an opportunity here. Perhaps, while they were doing well in the graphics workstation market, they should’ve produced graphic cards for other markets as well?
An SGI vid card probably could have competed with the likes of NVIDIA. They could have used their MIPS division some other internal hardware unit to come up with the final product. At the time they owned Alias, and was working on OpenGL, my stars that would have been a perfect showcase for an SGI card! They could have produced cards for high-end workstations to gaming stations without going ballistic with the price.
And if not the cards, then licence the tech or outsource the production of these things. Either way, it would have another revenue stream by now. Hell, maybe they could’ve become another source for GPUs or display chipsets for consoles or other embedded multimedia devices (like HD Tvs).
that was a bad move, also. If they don’t manage to turn it around, I’ll have to chalk up SGI as one of the 2 great tragedies of the computing world in my adult life to date – the first being the demise of DEC.
While everyone wants to blame Linux, but SGI was a shell of a company before it ever got involved with Linux.
What killed SGI was there dealings with Microsoft. They sank millions into porting NT to MIPS, when everyone was saying NT was going to rule the world, and Digital was doing the same porting NT to Alpha. In the end they both ran NT but no software makers ported there code to the platform. The real UNIX people got pissed, and moved on to other workstations and servers.
SGI’s management didn’t notice how much they were allienating there market, then the other shoe dropped with PC’s and there graphical accelerators. That did 80% of what SGI’s top of the line graphics power houses did for 1/10th the price. This left SGI’s market being Live video production TV stations and some animation studios.
SGI has gone out there way to allienate there followers since dot bomb and still today. IRIX had a strong following but the only way to use and develop for it legally is to pay $600 a year for a support contract. In the end no one is developing for it, so no cool software is being made for it. They should of given IRIX and compilers away free for developers, at least they could of had more settop boxes running MIPs and a semblance of workstation market.
There are no cool technologies left in SGI that are worth buying. Want lots of CPUs talk to Cray that own most of the great patents anyway. Cross bar curcuitry is awesome and came from cray unless i’m mistaken, but AMD’s hypertransport is good enough. Nvidia and ATI have exceeded SGI’s ideas long ago, and done it cheaper. No need for buy Tram (texture ram) that costs hundreds of dollars each. XFS was donated to Linux. All there super computer technology has been ported to linux.
The final hit, was SGI’s managment deciding to go with the itanium, the all but dead processor that will never have more than 5% of the workstation/server market.
Whose next? HP is next, they have made a number of marketing blunders over the last 5 years. No one is developing for HPUX any more, its only a matter time of before there workstation and unix server markets get spun off and HP focuses on its profitable Printer and x86 desktop machines.
>What killed SGI was there dealings with Microsoft.
No, not at all. It was neither Microsoft nor Linux.
It was because they didn’t understand that the market had changed and they didn’t respond to it.
I know you are a linux fan and naturally want it to be Microsoft but that’s just not true.
The fact is that Itanium or no Itanium, Linux or Windows, SGI would have died because they didn’t understand the change in the market and change when they needed to. They could have been like ATI or Nvidia now, but they are not.