Slashdot Mirror
SGI Steps out of the Visual Workstation Market
Lars Bergstrom writes "Well, SGI's finally given up on their Visual Workstation product line -- check here for the details. " As many people have noted, the technology was pretty sweet, but people won't pay the huge premiums for that. At least the flat panels are great.
Mind you, from all the VARs and such I know and hear complain, margins on hardware are slim going on nonexistant right now for just about everyone aside from maybe Intel and... uh... Intel.
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
"SGI will try to give Linux a boost by releasing some of its OpenGL graphics software into the open-source community, Stedman said."
...
Like GPL-ing it? While MesaGL is a worthy alternative, it would be in SGI's best interests to get OpenGL into wider usage. I suppose there may be licensing implications if they were to try and GPL the whole shebang, as many companies have already *purchased* licenses. So perhaps we'll see an open source 'official' OpenGL when those licenses expire
Chris Wareham
"Sexiest hardware in existance ..."
Personally I've got a soft spot for Sun pizza style cases, and improbably huge 24bit framebuffer cards. Must be a nostalgia thing I suppose. Although someone claimed to be able to get hold of recent Ultra workstations in Pizza boxes instead of those boring mini-towers.
Chris Wareham
The SGI investor page says it all
www.sgi.com/investors
(press reload a few time for maximum effect)
--------------------------------
I'm surprised people here are so ignorant about the realities of the marketplace.
... the XIO bus still beats the wet dreams of FutureIO and is here now) but then they do certify and guarentee the performance of the ones that do make the effort. If SGI was smart, they'd license the XIO technology to foster a vibrant alternative.
Facts
1) If a company doesn't have a competitive advantage, then sooner or later it will lose marketshare. The NT based Intel line (idea of the previous CEO) was targetting a niche which has been eroded by Dell's just-in-time assembly and creating a gap until HDTV standards become mandatory. SGI is being smart cutting their losses when they can. Unlike the disposable mentality that Wintel is trying to foster, SGI machines are designed to last for a long time (10 years+) which if your amortise the higher upfront costs, is actually cheaper in the long run.
2) SGI still has 300 odd engineers working on Irix and a growing pool of Linux developers to progressively shift their core competencies (multiprocessor design, low-latency memory subsystems) into value-added components. 10% of the market is seriously interested in high performance vs good enough.
3) Processors take a long time to die. Their MIPS line still has got excellent cost per application performance and decent data bandwidth (running a gigahertz Intel chip with IDE drives to surf the internet sounds a little fishy). As an embedded processor, the x86 is dwarfed by sales of MIPS cores cross-licensed to multiple vendors and their high-end stuff (R16000, etc) goes out to another 7 years.
4) The world market has been very very lucky in the Asian crisis has depressed global supply to such an extent that firms are willing to sell subsystems at near cost just to keep the cashflow up. In my opinion this has distorted the market and once things start bouncing back (as with the recent hike in memory prices) then their offerings will be more competitive.
5) Intel is becoming more like a venture capitalist than chip maker by throwing advanced designs at anything that looks, walks and talks like a high-growth startup. SGI is sorta doing the same thing by developing high-end, then flowing the tech to NDIVA (graphics), Cray (memory subsystem) and MIPS (CPU design).
6) Forget hardware, the value is now concentrated on the services. Even IBM are finding it difficult to move their big iron. Ultimately you will see companies becoming coordinators assembling components and software bundles to targetting specific market categories and capturing benefits from after-sales support and in the life-time money stream. You'd might be surprised how little car manufacturers make from sell the initial wheels compared with gouring consumers with repairs and add on knic-knacs.
7) Linux is shifting the competitive landscape in that every processors has an equal chance as the source is available to all players. Intel is recognising this and is dumping money to maintain dominance. If IBM can offer a PPC reference design, then perhaps SGI could do something similar for their MIPS, OpenGL and other graphics libraries (ie OpenAPI but retain implementation IP). Don't rule out the Koreas with their Alpha licensing and the Japanese with MIPS variants (e.g. Sony Playstation2).
8) Lighten up, markets go up and down in the short term. If anyone recalls the dumping Apple was going through several years ago, SGI is going through a similar patch. With a new CEO on board and some good Linux buzz, they have opportunities to catch the next wave. From a professional point of view, it would be interesting to see if their boxes could be adapted to keep the same fast memory subsystem but accept MIPS, Cray, Intel or Alpha processor node cards and just absolutely dominate the SMP and Beowulf market.
9) High end graphics is tough with very few firms having the capability to tackle complex real-time graphics and simulation. Today's multiplayer games are a shadow of hosting real-time thousands of players. If SGI could shift some of their talent to exploit the gaming niche, then they could gain more revenue streams from selling their IP. Admittedly they do have problems with delayed supply of third party boards (sigh
In short, SGI may be down at the moment but certainly not out. You don't have to be big so long as you can be profitable in your own little market niche, afterall BMW and Rolls-Royce still survive despite the Fords, Hondas, etc of the world. If people stop fixating on pure clock speed and look at real-world performance (e.g. broadband systems) they might be pleasantly surprised.
LL
Just wanted to make that clear.
CmdrTaco: was that headline your idea?
You are certainly becoming a journalist by all measures....
Dag B
Where did you read that? They are dropping some lines of workstations,
not "stepping out of the market". Their MIPS roadmap extends, it says,
to 2006 and their Intel plans are very much in place.
Slashdot could do with editors who actually read the subject matter.
I think it's important to note that their MIPS line of workstations, such as the O2, Octane, etc is still being sold. They're out of the NT workstation biz, but not out of the workstation biz period.
About a year ago when they introduced the NT line, I predicted that they would be unable to sell into what is largely a commodity market. Clearly, I was right.
Personally, I think they should abandon their plans to abandon Irix, which makes their high-end hardware feel distinctly orphan-ish. Either do that, or port Linux to the old machines so they won't become obsolete quickly.
Of course I'm still using my circa 1994 Indigo2, and I still love it. If SGI depreciation in the marketplace continues at this pace, I'll be able to afford a dual-processor Octane soon. But I'll cry for the company as I sign the check.
D
PS Happy Thanksgiving, Slashdot!
----
Unfortunately, this isn't a substantial profit centre for SGI, as they probably only make a few bucks per farmed-out MIPS CPU. To turn this into significant help, there need to be millions of CPUs being sold for this purpose, and I just don't see that happening.
This looked like a useful synergy; Cray didn't have the quantity of sales to allow development of all the hardware they needed, and if SGI could use some of it in other product lines, that could make the costs more readily amortized, and even improve performance on SGI's "own" product lines.
I suspect Cray holds a bunch of critical patents on computer hardware for HPC; killer question is who's going to want to buy, when SGI couldn't make the synergies work...
If you're not part of the solution, you're part of the precipitate.
according to some people I know, SGI will come out whit a new line of machines early next year. They will be linux only and have the same great IO as the NT machines possibly whit, firewire and USB, they will feature NVIDIA 3D graphics chip set (most likely some type of Gforce) Since it will be a linux only machine It doesn't have to be a x86 processor, it may even be a risc.
I think it is fairly likely that we will see some thing like the above, and I even think it may save SGI if the pricing is right. SGI is the best when it comes to IO, and whit the help of Nvidia and linux they could survive.
SGI came to be a hardware company because they needed specialised hardware to run their very cool graphics software. Their hardware did amazing stuff and people who did 3D graphics were willing to buy SGIs. Many cars were designed on SGI's. Even a little custom machine shop back home had an SGI running Engineering Pro. But that was yesterday.
Today the little machine shop is running AutoCad on a cheap but overpowered Intel box. A local 3D animation studio is using dual P2s with 3DMax. They looked at an SGI. It was sweet. It was faster. But not 6 times faster and that is how much more it cost. The Wintel stuff is dirt cheap and quite powerfull. SGI dosen't enjoy the nearly same performance magin that it did less then 5 years ago. The auto makers may still be buying the SGI stuff, but for how long?
The more we hear from SGI, the clearer it becomes that massive and fundamental changes have taken place inside the company, not just in public policy but in their understanding of their market and future prospects. Corporations of that size don't change direction on a whim and at a moment's notice. Like an oil tanker, there is a delay between cause and effect, so changes have to be well considered.
I don't have any involvement with SGI, but from the outside it seems to me that they've reached the following conclusions: (i) only their bigger machines are sector leaders and possibly still make a profit; (ii) their earlier preeminence in workstation graphics has been decimated by the collosal improvement in the capabilities of PC graphics cards; (iii) workstation-class CPUs and large memories are now commodity items, so SGI workstations can no longer claim that niche; (iv) SGI have excellent hardware techies but they cannot compete in this new commodity market because margins are far too small; (v) it is very difficult to compete against free software / open source in a market like theirs where users are technically competent, and "if you can't beat them, join them"; (vi) not all is well with the Microsoft titanic, and seats in the lifeboats are starting to look inviting so they are playing down their involvement with NT; (vii) in contrast, things look very rosy for Linux, in particular the wide acceptance among developers that this is A Good Platform, so they're playing that up to be developer-friendly and clearly "with it"; (viii) the synergy among other relevant corporates is massive in this area, and no way can SGI afford not to be on the same boat as, for example, Oracle; (ix) no major computer manufacturer has yet capitalized on the potential of Linux (nor the BSDs), and SGI could be The One that makes it "their own" if they genuinely adopt the ideals of the community; and finally, (x) software development (particularly maintenance) is incredibly expensive to provide, so making the most of open development on the Internet makes huge economic sense, ie. their overheads in that area could plummet.
All this adds up to a major shift, both internal and external: bye bye to the low-end proprietary stuff as per the announcement, leave computer basics to the commodity suppliers, at most customize PC workstations with high-end accelerators where margins still exist, leave Irix to their big systems where development costs won't decimate the spreadsheet, develop synergy with the free/OSS community, both as excellent value-for-money PR and as an essential component of their open support strategy, assist technically to give Linux some of the scalability that Irix has on their bigger machines, add in enterprise-critical facilities like a journalling filesystem, and in general hype up the whole scene so that they register in the books of PHBs as potential customers and shareholders.
The only thing that isn't consistent with this is their statement about continuing to support their MIPs systems at least until 2006, but I suppose that can be put down to not wanting to abandon their old customer base. That makes sense, as long as they don't spend too much money in this area of diminishing returns.
All in all, if their thinking is anything like the above then I reckon they have a good chance of making a success of it. They certainly can't be accused of doing nothing as the wind changes direction.
"The question of whether machines can think is no more interesting than [] whether submarines can swim" - Dijkstra
As far as I'm concerned, SGI bowed out of the workstation market when they began offering intel CPU's running Windows. Like Packard Bell.
I wish I had a nickel for every time someone said "Information wants to be free".
These are my friends, See how they glisten. See this one shine, how he smiles in the light.
One thing both companies have in common is a large installed base of loyal users that aren't cracking the wallets open as often as they "should" to upgrade. Both SGI and Apple have made some great machines, and they typically have a much longer lifespan than your typical PC, and people tend to hang onto them longer.
The genius of the iMac and the G3 is that it got many of the old Performa and Quadra users to buy a new Apple. SGI needs a similar product for their installed base, and the NT/Intel workstation wasn't it.
--
Business. Numbers. Money. People. Computer World.
No, I think he has a point. Sun has done well, but they've been running for the high end as fast as they can. They're ditching the workstation and low-end server market, and are covering their Sparc bet with full support for the IA64 commodity platform.
Sun and SGI can survive sitting on the high end (look at IBM and it's insanely profitable mainframe line), but eventually they're going to get cut off from the mainstream user base, and Solaris/Irix are going to get pegged as exotic and expensive server operating systems (much like OS/400 has). People will only want to use it if they absolutely have to.
In the long run, they're losing mindshare which will lead to the loss of marketshare. (Look at people who try to implement NT in midrange situations just because NT is a familiar quantity.) The fact that a commodity Dell box running Linux|NT|Solaris is even comparable with a custom Sun or SGI box is bad news in the long run for custom UNIX hardware.
--
Business. Numbers. Money. People. Computer World.
At that price I'd consider getting them shipped over (my girlfriend works for an international courier, so it wouldn't be too expensive :-) Where are you getting them from?
"The invisible and the non-existent look very much alike." -- Delos B. McKown
I am typing this on a loaded SGI 320.
When it debuted, it was a very good all around performar, and it had the highest fill rate of any intel based system.
Now, an Nvidia GeForce is just plain superior in almost every aspect. Higher fill rate, even in high res, 32 bit, trilinear modes. Faster, more capable geometry acceleration.
Any remaining areas of SGI superiority are probably due to driver optimization rather than hardware architecture. Nvidia hasn't had much call to optimize stippled lines, for instance.
The super-memory-system wasn't all it was touted to be. It worked well for sharing the load between the graphics and the cpu, but the cpu didn't actually see any better bandwidth than on a standard intel chipset. The cpu write bandwidth was actually about 10% LOWER than a consumer machine.
I have used a lot of intergraph and sgi machines, and the bottom line is that the consumer hardware has just outpaced the workstation hardware because they were on different growth curves. The workstations are better than they have ever been, but the consumer systems are just orders of magnitude better than they used to be.
I think SGI shot too low with the VW's graphics, somewhat out of fear of canibalizing their other workstations, and somewhat out of underestimating the consumer competition. Being quite a bit late didn't help, either.
John Carmack