There's probably young-uns here who don't know who Silicon Graphics is or what they did.
Silicon Graphics sold proprietary unix systems that were very heavily optimized for doing graphics rendering. Their workstations sold from $10,000 on the low end to hundreds of thousands of dollars. They were expensive, but if you wanted to do real time 3D stuff for animation, visual effects, CAD applications, or any type of graphics modelling, they were the only game in town, and you pretty much had to use their boxes.
Their products were amazing in their time, and they pushed the state of the art in computer graphics hardware for years. But, between the ability to create relatively cheap linux based render farms with commodity hardware on the server side, and ATI/NVidia creating great plug in graphics cards for PC's/Workstations, SGI wasn't able to compete. They slowly died and faded away into obsolescence. Some of the last remnants of their technology that people interact with frequently is the OpenGL api.
I spent most of grad school with one Octane on top of my desk (my primary machine) and a second one under my desk. Beautifully engineered, rock solid and great for their time. But SGI's steps post about 1998 were just crazy. Windows boxes with no software, a failure to realize how powerful linux systems were going to get, and losing their OpenGL team to NVidia.
Further tying the past to the present, Google now inhabits the offices that SGI had to leave behind as their business declined. As far as I can tell, Google kept the old building numbers, too, which explains why the first building Google moved into were numbered in the forties.
SGI killed themselves, just took a long time to die.
They had geniuses working in their graphics side, and when they decided some 15 years ago that "Unix was dead" and WinNT was the future, those geniuses left for Nvidia and other places. SGI never recovered and ended up reselling supercomputers with ATI cards in them.
Softimage for example - once they have done the port to WINNT and showed the world it can be done on that platform they have sold Softimage to AVID - which recently sold it to, where all good applications die, Autodesk (Microsoft in shadows)
After Microsoft bought softimage and delegated its development resources to porting SI|3D to WINNT and further architecture Sumatra (code name for XSI) to be NT only, there were only a few more blows to SGI that topled the castle.
Alias/Thomson division of SGI was merged with newly bought Wavefront into Alias|Wavefront - they've spent money on wavefront, and delegated AW to build a new platform, code name Maya, to compete with newly announced Sumatra. So, basically PowerAnimator and Explore products they already had lacked sales suddenly in awaiting of Maya (20,000$ list price + 20K for each of the modules, like artisan for example - really).
Then there was talks with Microsoft regarding OpenGL future - both of them tried to join powers and bring the world Fahrenheit... Microsoft though bought some small company making what was later known as DirectX.
Discreet (makers of high end compositing and editing applications) was bought by Autodesk and joined with their Kinetix group (formed by autodesk and yost group - makers of 3d studio) to consolidate new media division in Autodesk. Future plans were to migrate/and or/ develop new "systems" platform to linux. This materialized some years later, though not in full capacity. It showed its teeth when AVID strongarmed into PC market though.
With all this happening, SGI also introduced PC line of workstations with some innovative architecture designs (Cobalt) - which was ridiculously priced in heavy competition by then Intergraph Z workstations and Sun. There was a nice commercial tie in with movie lost in space - which, ironically was mostly made with 3dsmax (it's first feature credits)
ATI bought ART-X which made custom graphics chip for N64 among other things and Nvidia was hard pressed into buying talent around - mostly from advanced graphics division of SGI. This resulted in suffering in new lines of graphics workstations from SGI - like Octane2, Fuel and Tezro. Monster Reality engine was pretty much dead in the water by then - only brute force doubling of power.
And the cookie crumbled. SGI was almost fully vertically integrated hardware manufacturer - they had financial trouble. So they have ditched Alias|Wavefront as Alias to some investment fund (which refurbished it and sold it to Autodesk), MIPS division went from processor manufacturer to IP core provider only. R&D left to Nvidia, with appropriate patents (sold to Nvidia). PC division was dead. All that was left was HPC which sgi with its cray division focused on - unsuccessfully. Even though Altix platform is/was great - it was in direct collision with companies like Sun, RackSaver and most importantly IBM. IBM was still fully vertically integrated (almost still is) and had lots of money to throw around. It also had several strong HPC platforms (instead of one), and even grabbed a console processor market from MIPS/SGI in a feud before.
SGI is a sad story of bad decisions and bad timing. Had it kept media division and bought Discreet and Softimage instead of Cray, or even along with it. Had it scaled down MIPS to IP core only after lost war with IBM and reallocated resources better to align with Microsoft's vision of Talisman by keeping its R&D lab doing Talisman... Today we would have a company that would be a hybrid of a DCC giant and Nvidia. Now we only have a nice memory of turquoise and crimson wet dreams.
Oh they tried NT. Even played with one of those machines. They even looked ugly. They do a lot of Linux now, and have been playing with hybrid architectures. Another bet of theirs that failed were their Altix systems since the Itanium went nowhere fast
I know they tried to add NT to their portfolio, but they kept using IRIX for their supercomputing systems, as well as for their higher-end workstations until they switched to Linux for the supercomputers, and completely gave up on workstations.
Today their systems almost all use Linux.
Obviously, they didn't switch to NT, they switched to Linux.
the C++ STL library was based off of the header based standard developed at SGI. So if you've used STL before in C++, you're building off the work that came out of SGI.
Damn, SGI is right there with Digital in the category of amazing companies that brought us amazing advances and didn't deserver the fate that befell them.
...looks like Sun might be the next member of that club...
I worked there as an intern back in 2000. I did chip circuit verification, and saw that they had roadmaps for their future products up to around 2006. The group I was in was designing server chips and their interconnects to make clusters of high performance super computers.
I recall that they knew Intel was making dual and quadcore chips that would kill what they were doing, but had no solution how to turn the ship around.
It's too bad. There were some solid engineers there, especially the ones that came from Cray.
Giving up IRIX for WinNT? Well. Death to the unbelievers I say. I never forgot the shock when they had those ads with ther PeeCees in the computer magazines. Up until that point SGI and SUN where kind of the good guys to me.
Now, how could they have done it differently? I'd say: build up the lead in gaming consoles and developer workstations they had ad the time. IRIX/MIPS had also some technical advantages over Linux-Clusters in some niches like military simulations. They could have become some sort of NVidia in our days. Does any one has a more qualified strategy / commentary about SGI's demise in the markets to offer? Thanks.
SGI clearly had the cutting edge graphics technology and hardware - In my mind, the 3D video cards killed them very quickly and made SGI obsolete. As you may recall, when the 3Dfx Voodoo's were hitting the market, SGI did nothing at all, thinking that they still had the market cornered. Then came ATI, Creative, Nvidia etc. Back then there was the delineation between 'gaming' video cards and 'workstation' video cards, the latter having the premium price. SGI could have positioned themselves in both areas, especially the workstation cards.
It was a classic case of a large, entrenched, slow moving company getting thwarted by startups. Too slow or unwilling to act / react...
Hindsight is 20/20, of course, but were I SGI, back when PCs had just started getting to the point of running true 3D games, I would have had the company enter the graphics-card market. The SGI name could have easily carried a premium price-tag, and could have provided a compelling upgrade line for customers when PCs just didn't cut the mustard any longer.
This would have also let SGI position themselves to sell Maya across-the-board, giving them yet another solid revenue stream, and could have also let them jump into the console market, becoming the graphics vendor that powered the PlayStation, Sega Genesis, and so on.
Silicon Graphics sold proprietary unix systems that were very heavily optimized for doing graphics rendering. Their workstations sold from $10,000 on the low end to hundreds of thousands of dollars. They were expensive, but if you wanted to do real time 3D stuff for animation, visual effects, CAD applications, or any type of graphics modelling, they were the only game in town, and you pretty much had to use their boxes.
Their products were amazing in their time, and they pushed the state of the art in computer graphics hardware for years. But, between the ability to create relatively cheap linux based render farms with commodity hardware on the server side, and ATI/NVidia creating great plug in graphics cards for PC's/Workstations, SGI wasn't able to compete. They slowly died and faded away into obsolescence. Some of the last remnants of their technology that people interact with frequently is the OpenGL api.
Silicon Graphics, you will be missed.