Submitted by “Silicon Graphics announced today that Dennis McKenna has been named chairman of the board, chief executive officer and president, effective immediately. McKenna succeeds Robert Bishop, who will remain on the board of directors and serve as vice chairman. “Dennis McKenna is a proven leader, with an established track record of driving positive results in difficult business environments. He is well balanced in strategy, business development and operational execution – a combination that we believe will bring improved results to SGI and its stakeholders,” said James McDivitt, SGI’s lead director.” Update: El Reg has more on this.
Should have never abandoned the MIPS architecture. It is still superior in many ways to almost all current CPU architectrues. MIPS is what made SGI “High-end” now they are just another Itanium VAR.
No, the biggest bad decision was spinning off MIPS, which they used to own – atleast if they owned that, and everything went to the shitter, they still had the CPU busienss specialising in embedded processors to turn to.
The problem is, they have a new Itanium workstation, no commercial software for it, making it a VERY small niche – What they need is a big kick forward by focusing on the big markets, not the niches – that is where the money is.
When companies say, ‘concerntrate on our niche’ you know they’ve completely lost the plot and given up.
>>What they need is a big kick forward by focusing on the big markets, not the niches – that is where the money is.<<
Is that realistic for a company as tiny as SGI? If SGI is doing the same thing as everybody else, then who needs SGI?
If I were to consider buying from SGI, I would want to know: what different, and special, about SGI? What can SGI offer that I can’t get from Sun, IBM, or HP?
As it stands now, the reason to buy SGI, is to buy an old box to run legacy apps. At least that is the only thing I can think of.
Is that realistic for a company as tiny as SGI? If SGI is doing the same thing as everybody else, then who needs SGI?
For example, Itanium running Solaris (work with SUN to port it to Itanium, most of the work has already been done), thats a good idea, concerntrate on getting the desktop up to standard, and get applications on the platform by PAYING the companies to start providing them. Build up a network of ISVs, lower the cost of the workstations down to something dollar competitive with the Opterons being sold by SUN and voila, you’ll see volume make up for any lower margins.
For servers, same situation, they need to push their hardware not only for virtualisation but for also those who want to have a big friggin box for their Oracle database.
Its about using what is out there, and focusing on what they can do – so far they seem to have their head up their ass thinking that solely scientists are going to pull them out of the black hole they’ve thrown themselves into.
I totally agree with you. Time to see low cost MIPS64 desktops.
I’m wondering if anything can be done to save SGI. They are so deep into the Itanium architecture that they might not be able to get out of it. Even if the Itanium is “supposed” to become significantly better, IIRC it’s not going to come into fruitation until atleast late ’07 or ’08.
Is it normal to appoint a single person to all 3 top positions? I figured part of the reason there were so many was so that the power would be somewhat distributed.
It is not normal, but not unusual.
What it means in this case, is that SGI’s directors believe that they need a really strong man to get them out of this – so they are giving him every vehicle to do so.
History has shown that companies in trouble try to reduce the number of ‘parameters’ (impose stright strategies) – however, that often turns out to be the worst possible thing you should do.
SGI seems to continue researching and churning out new products at a fairly high pace – question is if they are being creative in the right or wrong areas.
Let’s see what happens.
CEO McKenna was confident about the future of the corporation and wanted to know if you were going to finish that sandwich.
“Should have never abandoned the MIPS architecture. It is still superior in many ways to almost all current CPU architectrues.”
I have nothing against MIPS per se, but by now people should have realised that its place is in the embedded market and nothing else.
The fastest MIPS boxes from SGI were just a mere 800MHz. Even the Alpha and the PA-RISC broke the 1GHz barrier… No, SGI saw MIPS as a dead end in the server/workstation space and chose another (better) processor to take its place.
The Itanium is great for SGI, it has great performance for the HPC crowd. Now, it never gained widespread use, and it is *very* expensive, which puts SGI in a bind…
I think SGI should think about selling Opteron-based servers. Both HPC servers (not everyone wants to buy an Altix mammoth, some just want to build cheap clusters) and middle-range servers.
At the same time that SGI is getting squeezed in the HPC space, others like HP are eating their cake as server suppliers for the graphics people (a traditional SGI market). Maybe those people would choose SGI Opteron boxes for their file-servers and renderfarm nodes if they just had them for sale…
Perhaps you’ve heard the term NUMA? While a single 800 MHz mips cpu isn’t anything to sneeze at, how about 512 of them in a single image sharing 1 Terabyte of memory? Some SGI machines are monsters of computing power.
Well, now imagine that same machine, but with Itaniums instead of MIPS (an Altix), and you have several times the computing power.
Keeping MIPS would have meant *huge* amounts of money to keep it a competitive CPU (performance-wise), and some more to keep IRIX from becoming a relic. And for what? What’s the difference in having a machine with a MIPS or Itanium CPU (or an Opteron, or an Alpha)? And what’s the difference of having IRIX or Linux?
In their day, SGI hardware with MIPS and IRIX was great, but times have changed. What does IRIX have that Linux doesn’t have right now?
You people are just being nostalgic. If SGI had kept MIPS and IRIX, their current products would be as capable as the Altix + Linux. Well, assuming they hadn’t go bankrupt a long time ago…
As I’ve said, the problem with SGI isn’t MIPS, or IRIX, or Itanium, or Linux, it’s their single-headed focus on the HPC market. The MIPS+IRIX to Itanium+Linux was a good move, but that doesn’t keep the HPC market from shrinking year after year.
Look at Sun, look how many Opteron boxes they are selling.
Edited 2006-02-02 01:07
Actually I’m of the opposite opinion. Why stick with mips and Irix? Because that’s what your customers applications run on. As it stands right now, anyone still left on mips will need to port all their existing applications to Linux on Itanium just to move to SGI’s latest offerings. Then consider how poor the performance of Itanium turned out to be, and add to that the fact that Itanium launched 2 years late or so forcing SGI to produce two more generations of cpus anyway. I’d say the move the Itanium was rather boneheaded. Maybe not as bad as SGI’s foray into NT machines, but still.
“CrLF”: I have no idea which statistics you are reading, but I believe you should stop smoking whatever you smoke while doing so and turn the paper the other way around.
You say that “but that doesn’t keep the HPC market from shrinking year after year.”
You are just dead wrong – the HPC market is growing at a hefty speed – and I know because I’m the CEO of a HPC company growing faster than that rate.
If you, however meant that “the number of players in the HPC market is shrinking” – you would be wrong as well. Although HP, IBM, and the other tier 1 players grow the market every year, there is a huge number of other players in the HPC market that sells cluster based storage solutions, cluster based visualization solutions, cluster based data mining solutions, cluster based almost everything else.
Make no mistake – the HPC market is thriving!
Edited 2006-02-02 04:24
You are just dead wrong – the HPC market is growing at a hefty speed – and I know because I’m the CEO of a HPC company growing faster than that rate.
If you, however meant that “the number of players in the HPC market is shrinking”
I didn’t say either. Well, I guess I made it fairly easy to misinterpret…
I meant HPC in the traditional sense. The need for high performance is still there and increasing, yes, but with expensive proprietary hardware? I don’t think so.
I really hoped that SGI would survive, but I really don’t see a way for them to do so this late in the game. I wish them the best, and hope they can pull a rabbit out of their hat.
So that’s why their shares jumped up.
But still, SGI can hardly be saved without diving into a new market. Itanium in retrospect seems like such a bad horse to be betting on, just like Windows NT on their machines was.
Anything they do seems to be done alternatively cheaper: via clusters, and their storage offers are about to be sidetracked via ZFS on cheap RAIDs.
The fastest MIPS boxes from SGI were just a mere 800MHz. Even the Alpha and the PA-RISC broke the 1GHz barrier… No, SGI saw MIPS as a dead end in the server/workstation space and chose another (better) processor to take its place.
So your dislike of MIPS technology is based on the Mghz Myth?….thats funny. They should have invested some R&D in to the platform and advanced it.
SGI should have kept Linux at arm’s length like IBM, Sun and HP. They should have said Linux is supported but never have made the fatal mistake of replacing IRIX with Linux.
Linux has turned out to be a kind of a pacman – devouring companies that stand to either partner or compete in the UNIX arena.
SGI didn’t replace IRIX, customers did!
It all happened at the beginning of the new (21st) century… when Intel’s PIII Coopermine hit 1GHz, and when NVidia announced GF2 – successor of the great “T&L wonder”. Then, recession hit the Blue planet, and suddenly price tag became inherent then ever.
When big Hollywood production houses began upgrading their workstations and render-farms they start looking at more costworth/scalable x86 alternatives with NVidia graphics. Just to be clear… SGI’s Wintel series of workstations were just about responding to that demand and not reinventing the wheel.
MIPS/RISC CPUs were great but they couldn’t oppose to laws of economy… Accompanied with fast networks and cluster/software engineering development, x86 took over allmost everything but embedded, mainframe and mission critical applications. Its production scale made it so cheap – it was to hard to resist, despite all its shortcomings.
SGI never bothered to port IRIX to x86 platform, thus making it obsolete. But those costly animators and FX experts didn’t want to depart from their trustworthy applications and production pipelines to which they are accustomed, so IT managers and administrators start looking for IRIX alternatives in order to protect their investments in staff tuition and propriatery (in-house) software. Naturally it was GNU, and Linux [u]KERNEL[/u] was just a popular open source kernel, which licence fitted their needs. That’s how Linux got involved in whole FX production bussines. LAMP servers and linux cluster market wasn’t SGI’s focus back then.
It’s a bit simplified and rough explanation but in essence that was the whole point of moving to Linux.
If SGI had responded to market demands more agile, they would still have some market share and would have reason and resources to port IRIX to x86, just like Sun did with Solaris. But that’s a lot of ifs… IRIX is dead now, and if it will ever comes back, I suppose it will be in form of compatability (API) layer on top of LSB, OpenSolaris or some other OpenSource system, with couple of development utilities and custom wallpapers.