Slashdot Mirror
Is the Game Finally up for SGI?
Rob writes to mention a Computer Business Review article looking at the bankruptcy of SGI, and whether the company is planning on a comeback. CEO Dennis McKenna is emphatic that the company isn't just looking for an exit strategy, but it's hard to see where they could go from here. From the article: "SGI has more challenges ahead, and I still find it hard to believe that after all of the chances it has had to run a profitable server and visualization business in the past it can miraculously do so now, selling lower-end boxes on even slimmer margins. But I'm hoping that the Chapter 11 has provided the necessary wake-up call for the company to get really lean really fast, because only from a more stable financial footing does it have any hope of fighting its way back onto new technology buyers' wish-lists."
You would think that the SGI name has enough high end appeal that nVidia or ATI would want to market SGI branded video cards. SGI could certainly be had cheap.
an ill wind that blows no good
You know, when you bring something back from the dead, it's never quite the same again, and you usually wish you hadn't. Let the company die while people still have fond memories of the brand, I say.
Oh no... it's the future.
They did just break a memory bandwidth record yesterday.
= 55
http://biz.yahoo.com/prnews/060717/sfm024.html?.v
--------========+++Dont Feed The Lab Techs+++========--------
For more than a few years SGI has just been a big "game" to the execs and the board of directors who allowed the charade and obfuscation to go on.
SGI cancelled their annual shareholder meeting in December.
They barely gave us a conference call in January.. McKenna wouldn't say anything.. And they've cancelled every call after that.
I sold my last shares long ago (except for one) and I hope they get sued into oblivion.
Apple should buy SGI (patents, know-how, take whatever they can from IRIX, OpenGL, etc) and kill off the rest of the stuff they dont need or cant sell.
We (The Dutch Weather Service) bought an Altix in April.
Their hardware rocks. The software - though complex, on three racks using a common file system - works.
I never believed in Itaniums, but for our code (heavy vectorizable, large memory models) they fly.
In short, if SGI collapses, in our market the loss will be quite noticeable.
The real problem here is how to do something that is different enough and desirable enough that people will pay more to buy it than the cost of making a mainstream box do it. Apple does this with an extremely well polished, well mannered software environment where everything "just works." There is a niche for that product class that won't be overtaken by Windows PCs anytime soon (or Linux PCs, for that matter.)
SGI's systems were well designed, but the problem was computing power increased to the point where the price/performance benefit of their boxes got too small to warrant serious consideration. Power became plentiful and cheap, and SGI's clients were Unix nerds so they could make other solutions work if they presented more cost effective alternatives. Even if those solutions were less elegant, they resulted in a better profit yield. In a free market that's enough to make the decision.
It's like that Dilbert cartoon segmenting customers - Smart customers are never a good bet. Of course that's exaggeration, but Apple appeals to those who want their computer to Stay Out Of The Way. That market segment is much less sensitive to hardware technology change, which is why Apple has lasted so long. Apple's customers don't WANT to be "smart" about computers, so they select a system that doesn't demand that. SGI's customers were high end power users - they were and are smart about computers. So when the technology changed, their users followed the changes.
I would like to see some smaller companies again push the limites of what we think of as "standard" computer designs, but as SGI has learned there is no money in such work and fabrication costs are prohibitive. The Lisp machines died out years ago, even more thoroughly. Maybe MOSIS and co will let someone get creative again, but for now the market seems to have decided, and the decision is for cheap and disposable.
"I object to doing things that computers can do." -- Olin Shivers, lispers.org
If OpenGL is going to be up for sale shouldn't the Free Software Foundation buy it? Otherwise couldn't Microsoft buy it and sit on it, preventing any real improvement on Direct3D's cross-platform competitor?
Why doesn't SGI just admit they're no longer needed to make computers, and just make graphics cards? Make the best OpenGL "accelerator" chips, give away lower-performance OpenGL libraries for free to keep the API popular and capture a new generation of developers. Sell some overpriced complete systems from mostly commodity HW to the high-end film and TV studios that need them. And release all the fancy extra tech accumulated over the years as plugins to apps actually successfully marketing them under other brands.
--
make install -not war
They effectively did this when they shed MIPS and the high-end graphics division. They may be designing their own system boards, but that's barely a shadow of what they used to do. They need to face up to the fact that they've lost whatever competitive advantages they had (workstations running their own high-end graphics hardware) and they can't compete against HP in the itanium server market.
I think their only hope would be to partner with nVidia and try to claw their way back into the visualization market, while they still have some reputation left. Some new high-end graphics hardware on a 2+2 itanium board might get them some attention, maybe enough to leverage a round of financing.
Just junk food for thought...
As noted elsewhere in this thread, an individual Altix machine is ludricously expensive (even for an Itanium machine). However, the interconnect on an Altix extremely capable -- up to 512 processors* in a single system image, very low latency. For certain applications, this -- the NUMA model -- is very big win. Whether putting Itaniums into this very nice interconnect is cost-effective really depends on your specific application: how much cache does it need, its computation to communication ratio.
- _Quinn
*: I guess this means a 1024 cores, now that Montecito's out.
Reality Maintenance Group, Silver City Construction Co., Ltd.
I have to laugh at comments like these as they can only be made by people who either have never used Sun hardware or software, or no longer keep up with the marketplace. Innovations that Sun continues to pump out (i.e. Sun x4600, x4500, Sol10, ZFS, DTrace, etc.) certainly deserve recognition and are highly useful. I use and admin all types of machines and OS's (mostly UNIX variants), and Sun certainly continues to be useful and relevant. In fact, they're better than most vendors in most areas.
But more and more the linux renderfarm world is taking over. Hey even Discreet is switching to IBM, Opterons and Linux. That is a day many thought would never come. I own a Origin 2000 and it rocks. I'm sorry to see SGI go. The Octane Rocked, hey even my O2 is a neat machine.
Man is the lowest-cost, 150-pound, nonlinear, all-purpose computer system which can be mass-produced by unskilled labor.
and IBM, and Sun. from a business perspective, IRIX and GL were simply ways for SGI to sell boxes, the same way OS X is "just" a way for Apple to sell boxes. grandparent needs to realize that stand-along OS or system software companies are the exception (Microsoft being the only really successful one, and that's largely due to a collection of other market forces; more commonly, they end up like SCO).
application software is somewhat different; there's a more viable market there. but SGI's not known for any of that; are they really likely to go head-to-head with folks like Adobe and Apple and win, starting nearly from scratch?
no, hardware remains the best bet for SGI to recover, if there's any way at all to pull it off.
i speak for myself and those who like what i say.
I think their only hope would be to partner with nVidia and try to claw their way back into the visualization market, while they still have some reputation left. Some new high-end graphics hardware on a 2+2 itanium board might get them some attention, maybe enough to leverage a round of financing.
Exactly... But I think they should stay the heck away from itanium. It's going nowhere in the long term, and like you say, HP has them beat in that market anyway.
SGI should continue going on with their servers and huge visualization systems, but they should partner with both nVidia and AMD, and one of the premium PC motherboard manufacturers, and should market motherboards, graphics cards, and complete systems ala Alienware to the medium-high end PC desktop and Server market... The revenue would be used to prop the company back up.
There is more of a market for high end gaming rigs today than there ever was for workstations. They could easily do 200 million of business in a year (which is what AlienWare did last year). Give it a couple years to start, and it could happen--if they pushed compelling products at prices that aren't insane. They sort of missed the boat back when they started switching to Intel. They had the right idea, but poor follow through. They though that they could ship commodity hardware at their usual prices... And they did, but they didn't ship many. Whoever has been in charge of the SGI pricing schemes should be shot. Apple had the right idea, make it compelling and somewhat affordable, and people will buy it.
If SGI made a 2x2 AMD 64 system that could take gobs of ram and had high speed busses and memory systems, tweaked video cards, etc, they could make a killing, and they wouldn't even have to do anything but help design it and slap their badge on it. If SGI made an nVidia graphics card and an nVidia motherboard (with SoundStorm, please), I'd buy a dozen! Heck, I'd buy it if it were $70 more than the nearest competitor, probably even if it didn't have any clear benefit!
Constitutional rights may be respected, repealed, or modified; but they must never be ignored.
You know, I still don't get it: what is the point of "enterprise" graphics cards (i.e. FireGL and QuadroFX), aside from decent Linux OpenGL drivers? Is there some reason you can't use a normal gaming card for stuff like Maya, etc.?
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz