SGI, Once Mighty Graphics Giant, Gobbled Up For Pittance by Rackable Systems

In 2000-01, when I was supporting a government agency, I ended up supporting a ton of SGI workstations and servers. SGI made awesome graphics cards and servers which were THE number one choice for everything from computational chemistry to realtime 3D modeling. I helped setup the first Onyx 3400 server at the agency, a high end SGI server that was big iron in every sense of the words.

In those days I had an O2 workstation on my desktop as well as a RedHat Linux box.

The problem, as Mr. Wolfe correctly points out, is that commodity Intel chips began to rival the MIPS processors in power and the really amazing proprietary graphics software for SGI, like Maya and Houdini, was ported to Linux. The move from SGI to Linux workstations where I worked was beginning in 2001.

It is truly a shame that SGI processor and graphics development couldn't stay ahead of the low end. They really did have something special once upon a time. Building WinTel boxes was a huge mistake for SGI as well as the author notes.

There is still a market for big iron. IBM/Sun (one company now?) and HP still compete in that space. There is still a market for very high end graphics as well. SGI should have had the talent to compete. Unfortunately they didn't have the business skills to do so. That was obvious 10 years ago.

But you know, SGI and Sun were right about RISC. They kept up, indeed, they kept AHEAD of Intel for years, and they only spent maybe a 10th as much on R&D and chip development.

Intel and AMD have to have *armies* of chip-heads, and amazing theoreticians to get to where they are, much less to make any progress. SGI had symmetrical multi-processors in the late 80s, early 90s. And they ran pretty cool. Didn't need fans all over the place etc etc.

The market choosing x86 chips is kind of inexplicable, really.

Or VHS over Beta... it's all the same stuff really. (S)He with the biggest marketing team wins.

jacog almost hit the nail on the head here. Inferior technical solutions often win out over superior ones if they are well marketed. This is the hurdle Linux always faces in trying to overcome Windows; it doesn't have the deep pocketed marketing campaign.

I said almost because biggest doesn't always equate to best. IBM spent serious money and had a very large team behind OS/2, which was certainly a superior solution to Windows 3.1, early versions of Windows NT, or Windows '95. They had flashy commercials which were absolutely ineffective unless you happen to speak Czech or find nuns appealing spokespeople for IT solutions. Big team, bad campaign.

How important is marketing? Let's put it this way: How many of you knew that OS/2 is still around and being developed. It was sold to Serenity Systems and is called eComStation nowadays. They recently released eCS 2.0 release candidate 6a. eCS 2.0 could have been called OS/2 6.0 if not for the name change. Most people aren't even aware of it because of minimal marketing.

/me raises hand

I even tried to test the beta, except having an AMD64 system w/ SATA drives got in the way.