Slashdot Mirror
SGI Faces Another Reorganization
dewey writes "This article [from Yahoo! News] says that SGI is expected to announce another business restructuring next Tuesday. No details about the reorg yet, but the buzz is that the focus will be away from big customers buying high-end machines and more toward being better able to 'compete on the Web' -- whatever that means." Update: It may mean more Linux support; jho sent in a link to the new SGI Linux page. A ray of hope in SGI's otherwise gloomy future, perhaps?
Two questions:
What's the non-linear editing software like, and how does one find a copy of it?
How do you find a reasonably affordable 100mbps Ethernet card for an Indigo2? I thought most of them were $ 800 plus.
D
----
---
---
Jamie,
Your O2 is safe, at least I see no 'end of life' for it in the foreseeable future. Bug fixes and improvements come with continued IRIX 6.5 releases and I think the hardware can be upgraded to an R12K.
I think SGI already dipped in the PC graphics market years ago - well before anyone else. Unfortunately, I think they were too far ahead of the field at that time. There was little or no PC graphics industry and their openGL PC board was ditched (maybe never really released).
The potential NVidia 'partnership' factor is interesting though.
I think we're all best waiting until next week rather than speculating on what they plan. I'd agree that SGI have made many mistakes - but, hey, hind-sight is 20/20!
Don't forget Samba, too.
SGI has been dipping it's toes in open software for more than a year now - not including the times open software has been incorporated into IRIX in the past (ie tcsh and top). Learning open software development practices has been an ongoing process at SGI - IRIX 6.5's maintenance process has a remarkable resemblance to those of several open source projects. They're also doing quite a bit to establish SGI as a valued member of the community by making contributions to the open source base. SGI is one of the few vendors who have realized that open source projects are community efforts and success depends on having the recognition and support of the existing community.
Note that effort to have Linux running well on the visual workstations have been redoubled, it looks like Fahrenheit was canned, and the new
Intel based low-end server is being pushed with more emphasis on Linux than on NT. Maybe a unix vendor found it hard to work with folks from Seattle. What a shocker!
Unlike the other major vendors that are hopping onto the open source train, SGI still has some remainder of the freewheeling, motivated by coolness factor and pride in achievement (that has to be flagrantly displayed) corporate culture that made them a leader in the first half of the decade. The same motivations that drive open source. The 'g' stands for geeks, maybe?
Another interesting thought is that while IRIX is certainly one of the most advanced OS's around, the only reason it exists is that SGI wants to sell hardware, but needs to have an OS that will actually take advantage of it. All that really takes is being able to contribute a few key pieces. They are bound to IRIX until another OS has it's capabilities but don't feel like waiting for a potential option to catch up - especially when they might make money supporting it now.
Is this /.? Or bad-attitude?
There's definately a move to put more diverse products into the market at SGI. And they've put good stuff out the door, but haven't been able to sell it in the volume necessary to pull a decent margin.
So how about 10) A bunch of people who played marketing experts on an Indy Cam find themselves unemployed on Tuesday and end up at Microsoft. A few years later, Linus achieves global domination. Even better, the engineers bought off by NetApp are revealed to have been working for satanic.org, a super-secret operations group of the ILUG. Netapp falls to (Linux MIPS based) Cobalt, and the penguin government purchases large numbers of big Onyx3's to model plans for paving Canada in realtime 3D. Said marketing experts are branded with a g and are exiled to Seattle without umbrellas.
Now is that b-a or g-a?
One would think that anything that SGI GPLs will be credited to them, and creditors might justifiably claim it was a fire-sale move and get the technology transfer declared illegal.
That's a really good point I hadn't considered.
---
DNA just wants to be free...
And many aren't. I work at a DOE lab, and we definitely do not buy enough SGIs to make their consumer business insignificant by comparison. Also keep in mind that the NSA has their own chip fab facilities, and those fables acres of underground computers are probably not made by SGI.
I agree with the person who said SGI should extend into the consumer graphics hardware business. I can only imagine how fast Diamond Mm et al. would fall to the might of SGI hardware.
*You* can't build an SGI system. That's the point. You are buying into SGI's expertise (in server-land) for things like fast buses, CC NUMA, scalability etc. This is the future - potentially. If SGI manage to get openGL running accelerated/optimised on their 320/540, that's another reason to prefer an SGI over one of your PC's.
It is open source and well maintained ;
http://www.CSUA.Berkeley.EDU/~gam3/knfsd/
I don't know what you mean by that last sentence.
Not only do I want references to the 'forward looking' analysts, but wonder why I'd trust them.
Maybe you work for Sun?
You ended with 'Sticks and Stones may break my bones but FUD will never concern me' but are guilty of spreading FUD yourself.
SGI have never announced the death of IRIX/MIPS. As long as MIPS competes, then IRIX will be around. What appears to be in doubt is the port of IRIX to IA64 - but even this is rumour and hearsay.
To a degree, it doesn't make sense to spend LOTS of money porting IRIX to Intel. If they're building Intel systems, it makes business sense to use Linux.
STOP SPREADING FUD
Yes, I do remember that roadmap. It very clearly delinated IRIX as living on IA-64 (Merced), and MIPS for quite some time to come.
NT was still at the low-end of the division, and was in no way replacing IRIX.
MIPS was the only part of the equation set to 'go away.'
And I just can't understand anyone who is "rankled" by this. Am I the only one with any sort of vendor loyalty who just doesn't care about NT?
Suffice to say, it's moot anyway. The big plan now is to add Linux to their roadmap, while IRIX remains on MIPS through R14000 and posisbly the R16000.
In my opinion, that would be suicide. SGI are not cut out to be 'graphics board' manufacturers. They build complete systems.
but if Roblimo's description is accurate, I don't think much of the approach. With SGI's NT efforts having a relatively poor reception, I would think they'd want to emphasize supercomputers and high-end workstations and servers as their core strengths.
D
----
Correct Link There was a ' instead of a ?
As callous as this sounds, I do hope they manage to get a good amount of their IRIX tech into Linux before they go belly-up. I doubt their creditors would be as free with it.
---
DNA just wants to be free...
I would have to agree. The high-end graphics systems are where it's at for SGI as they've really made their market there and have been able to dominate successfully. Maybe they see cheap distributed computing as a nail in their coffin and want to go on to something that would mean selling more units at lower cost. Still, there's a lot more competition in that market.
ian.
ian
One thing worth pondering, in the light of the serious period companies with commercial Unices like HP and SGI are now facing, is wether the success of Linux is what's killing the companies. People don't have to go for a big bucks system anymore like a decade ago. They can throw Linux on a PII system for pennies on the dollar.
Linux certainly has changed the Unix market. It might even have hastened the death of Unix (as a viable commercial platform).
Shame really - SGI have some really sweet technology. Been rather letdown by management lately it seems. I know some people who are kinda traditional buyers of high-end (ie Origin 2000s) SGI kit and they aren't too impressed with SGI's moves recently.
If SGI are going to drop/lose/sell some of their high-end stuff, I wonder who's going to buy... Sun bought the physical design for the Starfire about 3 years ago for $100M. Now they're making about as much revenue off the Starfire (including attached storage and services) as SGI are making in total...
Check out this C|Net article for more information
It describes how SGI is laying off and transferring a bunch of the Advanced System Division engineers. ASD is the heart and soul of SGI, and has been for a decade. It will be interesting to see what they say next Tuesday, but their actions are pretty revealing. Many engineers are being transfered to NVidia...recall that SGI reached some sort of deal with NVidia in the last couple of weeks about intellectual property.
The article says that the changes are designed to help their more profitable Intel workstations instead of their older Mips-based machines. Now, this of course flies in the face of reality, that the Intel machines lost a ton of money, but the Mips machines made enough that the entire company turned a profit. Of course, reality is a crutch for those who are not cut out to be Marketing Managers.
The article also confirms that the Fahrenheit initiative is being cut back. This is tremendously good news. Fahrenheit was supposed to be a follow-on to OpenGL, Inventor, and Performer -- it is a joint venture with Microsoft. I cannot imagine any input that Microsoft could have on OpenGL to make it better; even without the wretched example of Direct3D. If Fahrenheit was just a bone thrown to Microsoft to distract them from attacking OpenGL -- as it appears to me -- then I give SGI a lot of credit.
In articles a month or so ago, the announcement of the reorganization was going to happen today, the 5th, instead of the 10th. Their stockholder meeting, which used to be in August when I worked there, is now Oct. 27th.
Read the above article mentioned C|Net article, it's chock full of good information.
I love Mondays. On a Monday, anything is possible.
--
Please use the new logo, it's for you!
their pain.
But, on the other side, SGI has always had one
thing on its side: fan-fsking-tastic hardware.
And when the OS is the same, and the platform
is the same, what matters? Hardware. So maybe
they have a chance in the server market. If they
have fun with stupidly high bandwidth busses and
don't get greedy with the prices, they may do alright.
But for God's sake man, LEAVE THE HIGH END MARKET
AS IS! The world needs at least one vender doing
what SGI's doing. Sheesh.
--
Care about electronic freedom? Consider donating to the EFF!
I was thinking it might come with the video capture system for the box, which I've seen for sale on occasion.
D
----
You can generally get:
3COM 597 TX Controller
10/100 Base T Ethernet
Drivers available free on web
www.phobos.com
ftp://ftp.phobos.com/drivers/SGI/
For roughly $200
The $800 US Phobos G160 Fast Ethernet boards attach to the Indigo2's proprietary GIO64 bus which has much higher bandwidth than the EISA bus.
Unfortunately, this card wasn't much faster than the built-in 10BaseT Ethernet with the drivers available for IRIX 6.2. A very depressing realization considering that the card costs more than eight times that of a PC 10/100BaseT card. I just started upgrading to IRIX 6.5.4 and it looks like performance has improved quite a bit.
"Yeah well
No biggie on the spellcheck. The guy you were responding to spells definitely the new ignorant/geek way -- "definately." Someone notify Websters
Consigned to flames of woe.
The company suits decide "The world is moving to NT. We must move to NT."
The techs, of course, respond with "Oh god no." So they push Linux as an alternate hot, emergent technology -- one that doesn't suck.
The best parts of Irix technology can be moved into Linux, they can still make awesome hardware, the company does well, and everybody's happy.
--
The last major problem AFAIK is Linux's scalability to 4 proc and up. And we all know that this is a priority within the kernel bunch right now. By the time Linux 3.0 rolls around, we may have a very serious kernel on our hands.
--Lenny
look sgi should _not_ try to out-sun sun or out-ibm ibm. they have great technology in an area that could represent high growth in the next 5-10 years. they should move into consumer video cards and propogate technology through set-top and otherlike markets. don't be a follower, be a leader!!!
p.s. the linux strategy is really nice however...
-- your knees hurt, don't they?
Their project list on http://oss.sgi.com/projects/ lists the following ongoing projects:
--
The world is neither black nor white nor good nor evil, only many shades of CowboyNeal.
None of the vendors you mention offer what SGI does.
I suggest you examine SGI's website and consider the Infinite Reality Monster and the future potential of scaling to 512+ processors reason enough.
On this end, there are those of us who like running Alias (Maya), Lightwave, and Softimage: apps which don't run on SGI's ersatz competition.
The only way you can beat that system for 3D graphics performance is to spend $2000+ on a high-end graphics card such as the Intergraph Wildcard 4105. Of course you still have to buy a well configured PC to stick that card in. As far as 2D graphics goes the SGI VW 320/540 is unbeatable. Check out the review at Lumis3D.
If you're not doing highend 2D/3D graphics then don't buy the SGI VW320.
Admittedly, if you needed to do graphics work on the 320 today, you'd have to run WinNT, but it looks like SGI is committed to making Linux do 3D graphics well, in the very near future. There's gonna be some serious Linux goings on at SI99RAPH
"Yeah well
I think that the graphics are coming. There have got to be lots of legal issues regarding how they do graphics, and what they can or cannot release openly. The open source methodology may also slow them a bit. They just cannot rewrite what is there. That would not be good for those who have worked hard over the years. They are having to intergrate what they know, and sell it to those who are helping. If you have ever sat down at one of their workstations, it would become obvious that they know how to do X + OpenGL. The problem is that not everyone has been able to check out their technology. The only reason that NT happened so quickly is that they only had to Tweak OpenGL, and the underling HAL. Linux needs more work than this.
I hope they will be OK as well.
Blogging because I can...
for years they were the cream of the crop in servers and high speed workstations. Intel was never a match for them, but now with the PII/III and Xeon chips, Intel has found a way into the high-end computing market. I don't think SGI euipped itself to deal with this. They embraced Intel with the Visual Workstations, which are great pieces of hardware, but are too expensive for some, and can be seen as entering their dotage as even faster bus speeds are reached with non-SGI chipsets, 3.6 gig memory bandwidth isn't as fast as it was a year ago. It's good that they've embraced OSS because they've done some really great things that it would be nice to have open sourced but I think it's the beginning of the end of SGI unless they come up with the next Big Thing before someone else does.
I'm a loner Dottie, a Rebel.
I think SGI's best hope for survival would have been for them to have expanded into the PC graphics board business in 1993 or 1994. They should have been 3DFx -- instead, 3DFx was founded by SGIers who jumped ship. It's far too late for that now.
I remember being stunned at the marketing campaign for the Indy -- they had an ad where someone was saying, ``finally, SGI is making a computer that's just like everyone else's!'' Yeah, except that it cost 3x as much! They got the idea that they should be building these low-end general-purpose Unix workstations, and were actually downplaying the fact that these machines all came with kick-ass graphics as a standard feature.
It's hard to tell what business SGI is actually in these days, after all of these reorgs. They seem to have sold off the part of the company that does graphics hardware. What's left? Has SGI turned into a VA Research clone?
This is all very sad, because SGI has done some amazing things.
And you know, Sun can't build graphics hardware or software to save their life. Never have. It's sad to see that some people think that Sun is actually competition to SGI. Well, they are competition, but they're just not in the same league, dammit.
This means I'm never going to be able to get spare parts or bug fixes for my O2, doesn't it? Sigh.
At first when I read this and especially the rumor
that they are "downsizing" the team working on
a replacement for Infinite Reality graphics (now 4 years old) I thought that's it, the end for sgi.
-
But maybe now the whole thing makes sense. SGI had already lost a whole bunch of their hotshot
graphics engineers to Nvidia. So now they give the rest of their graphics engineers to Nvidia and outsource future graphics pipe development to them. Arguably, SGI hasn't been able to compete on graphics chip development for several years now anyway.
But their's one thing that noone else can do like them: Massive internal bandwidth. So they concentrate on developing radical new bus architectures and massive scalability (their core strength) and leave cpu development to intel (merced) and graphics chip development to Nvidia. SGI's almost certainly getting royalities from Nvidia anyway now that the patent swap thing has happened.
Not such a bad plan really.
Bus, bus, bus.
There's your reason. But, really, if you can build your own computer, why don't you just go and do it instead of whining about not being able to afford an SGI prebuild and warrantied for you?
Personally, I dislike building computers from scratch. But you don't see me whining about how much more I'd spend if only someone would od it for me, do you?
Bus, bus, bus.
There's your reason. But, really, if you can build your own computer, why don't you just go and do it instead of whining about not being able to afford an SGI prebuilt and warrantied for you?
Personally, I dislike building computers from scratch. But you don't see me whining about how much more I'd spend if only someone would do it for me, do you?
I really hope they get their act together, find the right direction and make it happen. Otherwise, they might face a similar fate like Digital, who also had great technology (Alpha chip) but no direction, and is now going down together with Compaq.
Under capitalism man exploits man. Under communism it's the other way around.