I remember having access to an Irix box at my first job. It was seen as a real, professional, serious OS, not like the Linux box I set up. Pretty incredible how that all changed in a matter of years.
A close friend of mine growing up had a parent that worked at SGI. I'm not trying to start a holy war, but I just have to let it be known that emacs was the recommended editor at SGI. Just saying. Maybe other things contributed to the fall but, in my heart I will always remember emacs.
Oh how I lusted over the Challenges, the Octanes, the Indigo2s of the time. It was a revelation when I finally was able to sit down at a console of an Octane (with two, count 'em TWO R14000 and a whopping 2.6G of RAM), tooling around in IRIX via 4dwm was so much more satisfying than today's UIs. It was snappy and low-latency unlike anything I've used since.
Later on, I was able to do some computational work on an Altix 3700 with 256 sockets and 512G of RAM spread over four full-height cabinets with the nest of NUMAlink cables at the back), at the time running SuSE linux and that was wild seeing the 256 sockets being printed out with a cat /proc/cpuinfo. Now the same capabilities are available in a 4U machine.
The corporate lineage story is also just as interesting as the hardware they made as well. Acquisition, spinoff, acquisition, rename, acqusition, shutter, now perhaps just a few books and binders and memories in the few remaining personnal at HPE are all that's left (via Cray, via Tera, via SGI, via Cray Research).
I still keep a maxed out Octane2 in running order for posterity. Occasionally logging in to it reminds me just how a desktop environment should feel. We truly have lost something since then.
I really wish they'd do movies like they made for RIM about: Cray, DEC, Compaq, SGi, Pixar. sounds like these places were either wild or strait up IBM culture or some clash or both either inside or outside. Raven and id Software would be neat too. Westwood Studios.
I remember having a Personal Iris, at the company I worked at, and, later, an Indigo. We never used them. I think they were really there, to impress the visitors (They were in our showroom).
I remember the colors as being very different, from the photos, though.
The Personal Iris was a deep purplish-brown, and the Indigo was ... indigo.
Jim Clark sounds like my kinda guy. I made a hash of my teenage years, and barely squeaked in, with a GED, myself. It has all worked out, OK, in the end, though.
When I was in high school we had a lab full of SGI machines. They also never got used. Hundreds of thousands of dollars of computing equipment, and probably that much again in software licenses (at the commercial rate), just sitting there doing nothing. It was heartbreaking.
On a happy note, the SGI bus (a semi-trailer full of SGI machines demoing their capabilities) came to school one time. As a teenage nerd, getting to play with a refrigerator-sized Onyx2 was a good day.
There were all kinds of toys, though. There was a dedicated classroom setup for video-based remote learning some 30 years before COVID - that got used for one semester, from what I gather (was never used while I was there). The school was even host to a dialup ISP at one point.
The administrators were all in on technology. The teachers, not so much...
Eventually, in my last year, the government changed the funding model and the party ended.
I worked on the SGI campus as a consultant/vendor to them in 1999/2000 during the dot.com boom. I really wanted one of those 1600SW flat screens (everything was CRT back then), but they weren't really in use at the time.
One of the neatest things is that they let us (Trilogy/pcOrder) put together a sand volleyball team to compete in their company intramurals.
That cafeteria went on to be known as "Charlie's" at Google and was the main HQ cafeteria (serving great food, and then later, extremely meh food). TGIF was also held there. If there ever was a place that was "Google central", that was it.
There's a few cases in the history of computers where it feels like the world just "chose wrong". One example is the Amiga; the Amiga really was better than anything Apple or Microsoft/IBM was doing at the time, but for market-force reasons that depress me, Commodore isn't the "Apple" of today.
Similarly, it feels like Silicon Graphics is a case where they really should have become more standard. Now, unlike Amiga, they were too expensive to catch on with regular consumers, but I feel like they should have become and stayed the "standard" for workstation computers.
Irix was a really cool OS, and 4Dwm was pretty nice to use and play with. It makes me sad that they beaten by Apple.
"Revolutionaries rarely get to live in the societies they created"
I think it's a combination of a skillset/culture needed to create a paradigm shift isn't the same one needed to compete with others on a playing field you built, and of complacency. It happens over an over. We saw it happen with RIM, and we're watching it happen right now with Prusa Research.
Both Prusa and SGI are (and were) probably largely unknown to 90% of their potential market. The globally recognized companies tend to spend far more on marketing than anyone in a STEM field would consider remotely reasonable.
> Similarly, it feels like Silicon Graphics is a case where they really should have become more standard. Now, unlike Amiga, they were too expensive to catch on with regular consumers, but I feel like they should have become and stayed the "standard" for workstation computers.
I think you highlighted very correctly there, though, why SGI lost. It turned out there were cheaper options, which while not on par with SGI workstations initially, just improved at a faster rate than SGI and eventually ended up with a much better cost/functionality profile. I feel like SGI just bet wrong. The article talks about how they acquired Cray, which were originally these awesome supercomputers. But it turned out supercomputers essentially got replaced by giant networks of much lower cost PCs.
Yeah, I'm more annoyed about Amiga than SGI. They were priced competitively with Apple and IBM offerings.
I guess it's just kind of impossible to predict the future. I don't think it's an incompetent decision to try and focus entirely on the workstation world; there are lots of businesses that make no attempt to market to consumers, and only market to large companies/organizations, since the way budgeting works with big companies is sort of categorically different than consumer budgets.
But you're absolutely right. Apple and Windows computers just kept getting better and better, faster and faster, and cheaper and cheaper, as did 3D modeling and video editing software for them. I mean, hell, as a 12 year old kid in 2003, I had both Lightwave 3D (student license) and Screenblast Movie Studio (now Vegas) running on my cheap, low-spec desktop computer, and it was running fast enough to be useful (at least for standard definition).
Of course, the reason they got better so fast is volume. There was just way more investment into those platforms. Which means this explanation is somewhat circular: they were successful because they were successful.
I think a more useful explanation is that people rate the value of avoiding vendor lockin extraordinarily high, to the extent that people will happily pick worse technology if there's at least two competing vendors to choose from. The IBM PCs were not good, but for convoluted legal reasons related to screwups by IBM their tech became a competitive ecosystem. Bad for IBM, good for everyone else. Their competitors did not make that "mistake" and so became less preferred.
Microsoft won for a while despite being single vendor because the alternative was UNIX, which was at least sorta multi-vendor at the OS level, except that portability between UNIXen was ropey at best in the 90s and of course you traded software lockin for hardware lockin; not really an improvement. Combined with the much more expensive hardware, lack of gaming and terrible UI toolkits (of which Microsoft was the undisputed master in the 90s) and then later Linux, and that was goodbye to them.
Of course after a decade of the Windows monopoly everyone was looking for a way out and settled on abusing an interactive document format, as it was the nearest thing lying around that was a non-Microsoft specific way to display UI. And browsers were also a competitive ecosystem so a double win. HTML based UIs totally sucked for the end users, but .... multi-vendor is worth more than nice UI, so, it wins.
See also how Android wiped out every other mobile OS except iOS (nobody cares much about lockin for mobile apps, the value of them is just not high enough).
What smaller businesses are using will tend to be what takes over in the future, just due to natural processes. When smaller businesses grow, they would generally prefer to fund the concurrent growth of existing vendors that they like using than they are to switch to the existing "industrial-grade" vendor.
At the same time, larger organizations that can afford to start with the industrial-grade vendors are only as loyal as they are locked in.
I mean, there are corporations who only sell to very large corporations and have had plenty of success doing so. Stuff like computational fluid dynamics software, for example, has a pretty-finite number of potential clients, and I don't think I could afford a license to ANSYS even if I wanted one [1], since it goes into the tens of thousands of dollars. I don't think there are a ton of startups using it.
But I think you're broadly right.
[1] Yes I know about OpenFOAM, I know I could use that if I really wanted.
I see the same trend in programming languages. Say a really solid career lasts from about 20 to 60, 40 years long. Say that halfway through your career, 20 years in, you're considered a respectable senior dev who gets to influence what languages companies hire for and build on.
So in 20 years in, the current batch of senior devs will be retiring, and the current noobies will have become senior devs.
*Whatever language is easy to learn today will be a big deal in 20 years*
That's how PHP, Python, and JavaScript won. Since JavaScript got so much money poured on it to make it fast, secure, easy, with a big ecosystem, I say JS (or at least TS) will still be a big deal in 20 years.
The latest batch of languages know this, and that's why there are no big minimal languages. Rust comes with a good package manager, unit tester, linter, self-updater, etc., because a language with friction for noobies will simply die off.
One might ask how we got stuck with the languages of script kiddies and custom animated mouse cursors for websites. There's no other way it could turn out, that's just how people learn languages.
Back in the old days there was a glut of crappy bloated slow software written in BASIC. JS is the BASIC of the 21st century: you can write good software in it, but the low bar to entry means sifting through a lot of dross too.
My take: that’s just fine. Tightly crafted code is not a lost art, and is in fact getting easier to write these days. You’re just not forced into scrabbling for every last byte and cpu cycle anymore just to get acceptable results.
This betting wrong on specialization happened over and over again in the late 70s and 80s. The wave of improvements and price reduction in commodity PC hardware was insane, especially from the late 80s onwards. From Lisp machines to specialized graphics/CAD workstations, to "home computer" microcomputer systems, they all were buried because they mistakenly bet against Moore's law and economies of scale.
In 91 I was a dedicated Atari ST user convinced of the superiority of the 68k architecture, running a UUCP node off my hacked up ST. By the end of 92 I had a grey-box 486 running early releases of Linux and that was that. I used to fantasize over the photos and screenshots of workstations in the pages of UnixWorld and similar magazines... But then I could just dress my cheap 486 up to act like one and it was great.
Atari ST and Intel PC are not distant categories. Both are "'home computer microcomputer' systems". Not all home computer systems can win, just like not all browsers can win, not all spreadsheets can win, not all ways of hooking up keyboards and mice to computers can win, ...
The people that created the Amiga weren't the same people as the ones leading Commodore. Apple's success seems to have been heavily based on the company's leader being very involved in product development and passionate about it.
Yeah, I think that would also have been a better timeline; I'm just stuck in the anglo-world and thus my knowledge is mostly limited to what was released in the US or Europe.
We've seen again and again that the high end of the computer market can't sustain itself; the mass market outruns it. The result is that the high end works best when leveraging the mass market instead of trying to compete with it.
See the dominance of Threadripper in workstations, which is built on top of mainstream desktop and server parts bin. Or look at the Epyc based supercomputers, rumored to be the only supercomputers to turn a net profit for the suppliers, thanks to leveraging a lot of existing IP.
It’s just a lesson in worse is (often) better. If you can do some most of the job with something that is either cheaper, easier to build, or easier to iterate on, then it will often overtake a better engineered solution.
My main problem with Silicon Graphics (& have the same problem with Sun Microsystems) is that they just tried to do too much in propriety hardware and completely resisted standards. Microsoft & IBM "won" because they made computers with actual upgrade paths and operating systems with wide support among upgrade paths. With SGI/Sun you were very much completely locked in to their hardware/software ecosystem and completely at the mercy of their pricing.
In this case, I think the market "chose right" - and the reason that the cheaper options won is because they were just better for the customer, better upgradability, better compatibility, and better competition among companies inside the ecosystems.
One of the most egregious things I point to when discussing SGI/Sun is how they were both so incredibly resistant to something as simple as the ATX/EATX standard for motherboard form factors. They just had to push their own form factors (which could vary widely from product to product) and allowed almost zero interoperability. This is just one small example but the attitude permeated both companies to the extent that it basically killed them.
> With SGI/Sun you were very much completely locked in to their hardware/software ecosystem and completely at the mercy of their pricing.
How is that in anyway different from Apple today with it's ARM SoCs, soldered SSDs and an OS that requires "entitlements" from Apple to "unlock" features and develop on?
You can buy a cheap Mac and easily write programs for it. You don't have to spend $40k on a computer, you don't have to buy a support contract, you don't have to buy developer tools.
Every time I’ve checked over the last decade (including today), you can buy a mac mini that supports the latest macOS for under $250 on ebay. You can also test your app using github actions for free if your use case fits in the free tier.
There is no way to do this for an IBM z16, which is the kind of vendor lock in that people are saying Apple doesn’t have.
The big exception here is that SGI took IrisGL and made it into OpenGL which as a standard lasted far longer than SGI. And OpenGL played a critical role preventing MSFT from taking over the 3D graphics market with Direct3D.
When I say "hardware graphics market" I'm referring to high performance graphics workstations, not gaming. There is a whole multibillion dollar market there (probably much smaller than games, but still quite significant). It's unclear what carmack's influence on the high performance graphics workstation environment is, because mini-GL left out all the details that mattered to high performance graphics (line rendering would be a good example).
In my opinion, Mesa played a more significant role because it first allowed people to port OpenGL software to run on software-only cheap systems running Linux, and later provided the framework for full OpenGL implementations coupled with hardware acceleration.
Of course, I still greatly enjoyed running Quake on Windows on my 3dfx card with OpenGL.
Well, put that way it is a market that runs on Windows with OpenGL/DirectX nowadays, or if using GNU/Linux, it is mostly with NVIDIA's proprietary drivers, specially when considering VFX reference platform.
If Amiga really "deserved" to win, I think they wouldn't have been eclipsed by the PC ecosystem in terms of performance.
They leapt out ahead of the competition with an advanced OS, purpose-built for graphics and sound in a way that PCs and Macs weren't.
Which was great. But they weren't really better than the competition. They were just doing something the competition wasn't. And when the competition actually started doing those things they got eclipsed in a hurry.
I wonder if Tesla will suffer the same fate. They were obviously around a decade ahead of the established players when it came to electric cars. But once the other established players actually got serious about electric cars, Tesla largely stopped being special, and upstarts like Lucid and Rivian are neck and neck with them (in terms of compelling products, not sales) as well.
"the future is already here, it just isn't evenly distributed."
This means there are products out there with futuristic features that will be seen as requirements for all things going forward and right now those features are niche elements of some product.
The Amiga was a fantastic device but not a general purpose device. Lots of things are fantastic at a niche but not general, and those almost always fail.
Yeah, and OS X more or less mainstream-ized consumer UNIX as well. It gave you access to the UNIX tools in the command line if you wanted them, had a solid UNIX core, but was a lot cheaper than an SGI and also easy to use.
Well, wait, the Amiga had preemptive multitasking way before Apple or Windows got it, like the mid 80s. I don't think Windows got it until Windows NT, and it didn't become mainstream until Windows 95. Macs had bizarre cooperative multitasking that would freeze if you just thought it about it funny [1] all the way until OS X.
There's other stuff too; they had better color graphics in the 80s while DOS was still dealing with CGA and EGA, and decent sound hardware. Even by 1990, the video toaster was released, well before it got any port to DOS.
[1] I'm sure it got better, my first exposure to it was System 7 and that thing was an unholy mess. I didn't touch macOS again until OS X.
Long before Windows 95 there was DOOM and DOOM would not run on an Amiga.
> 80s while DOS was still dealing with CGA and EGA, and decent sound hardware.
And then the 80s ended. What point did I make that you are contradicting?
> Even by 1990, the video toaster was released,
And if you wanted to do CAD? Would you use an Amiga? Probably not. What about desktop publishing? Pointing out that Amiga had carved out a niche (in video editing) when that was the norm back in those days doesn't make any strong comment about the long term superiority or viability of the platform.
Also, I don't buy into the idea that just because a company had something "superior" for a short period of time with no further company direction that they didn't lose fair and square. That Amiga had something cool in the 80s but didn't or couldn't evolve isn't because the market "chose wrong". Commodore as a company was such a piece of shit it made Apple of the 80s look well run. Suffering a few more years with the occasional bomb on System 7 was not a market failure.
> Macs had bizarre cooperative multitasking
What was bizarre about it, compared to any other cooperative multitasking system of the time?
Also you seem to be fixated on preemptive multitasking to the neglect of things like memory protection.
> Long before Windows 95 there was DOOM and DOOM would not run on an Amiga.
Yeah fair. I do wonder if a port like the SNES version would have been possible if id would have greenlit it, but that's a "what if" universe. Alien Breed 3D would run on a 1200, but IIRC it ran pretty poorly on that.
> And then the 80s ended. What point did I make that you are contradicting?
I mean, yes, VGA cards and Soundblaster cards were around in 1990, but they weren't really standard for several years later.
> And if you wanted to do CAD? Would you use an Amiga? Probably not. What about desktop publishing? Pointing out that Amiga had carved out a niche (in video editing) when that was the norm back in those days doesn't make any strong comment about the long term superiority or viability of the platform.
Also fair. I'll acknowledge my view is a bit myopic, since I don't really do CAD or desktop publishing, but I do some occasional video editing, and I do think Amigas were quite impressive on that front. You're right in saying it was a "niche" though.
> Commodore as a company was such a piece of shit it made Apple of the 80s look well run.
No argument here. Still think that the hardware was pretty cool though.
> What was bizarre about it
I guess "bizarre" was the wrong word. It was just really really unstable, and System 7 would constantly freeze for seemingly no reason and I hated it.
> Also you seem to be fixated on preemptive multitasking to the neglect of things like memory protection.
I feel like if Commodore had been competently run, they could have done work to get proper protected memory support, but again that's of course a "what if" universe that we can't really know for sure.
I guess what frustrates me is that it did genuinely feel like Commodore was really ahead of the curve. I think the fact that they had something pretty advanced like preemptive multitasking (edit: fixed typo) in the mid 80s was a solid core to build on, and I do kind of wish it had caught on and iterated. I see no reason why the Amiga couldn't have eventually gotten decent CAD and Desktop publishing software. I think Commodore didn't think they had to keep growing.
The Amiga OS was designed in a way that protected memory support was basically impossible. Message passing was used everywhere. How did it work? One process ("task", technically) sent a pointer to another, a small header with arbitrary data, which could contain anything, including other pointers. Processes would literally read and write each other's memory.
This is I think the premise that you and people like me who think Amiga could have gone on to do great things disagree on, I think. Most Amiga fans would say that it totally had a path forward, or at least there is no evidence that it didn't, and the failure to follow that path therefore it wasn't an inherent technical problem, but a problem of politics and management. Do you have any evidence to the contrary?
Commodore's story is more about achieving the impossible with 1-2 engineers building each computer. Commodore was a company built around Jack Tramiel who wanted his widgets to ship in volumes to "the masses, not classes". When he left then it was a lifestyle sucking cash machine for Irving Gould who appointed incompetent CEO after incompetent CEO after Tramiel. The miracle is it staggered on ten years post-Jack.
But the reality is the Commodore 64 kept Commodore going during most of that period rather than Amiga sales. It's similar to Apple where the Apple 2 kept Apple afloat during the 80s and 90s until Steve returned.
Times changed though, too, and Tramiel couldn't replicate his success w the C64 at Atari Corp, despite bringing the same philosophy (and many key engineers) over there.
By the late 80s the "microcomputer" hobby/games market was dead and systems like the ST and Amiga (or Acorn Archimedes, etc.) were anachronisms. You had to be a PC-compat or a Mac or a Unix workstation or you were dead. Commodore and Atari both tried to push themselves into that workstation tier by selling cheaper 68030 machines than Sun, etc, but without success.
>One example is the Amiga; the Amiga really was better than anything Apple or Microsoft/IBM was doing at the time
Amiga was only better 1985-1988.
I still have my original Amiga and A2000. I was an Amiga user for a decade. They were very good. I was platform agnostic, caring only to get work done as quickly and easily as possible so I was also an early Macintosh user as well as Sun and PA-RISC. And yes, I still have all of those dinosaurs too.
By 1987 PC and Mac caught up and never looked back.
But by 1988 the PS/2 with a 386 and VGA was out and the A2000 was shipping with a 7MHz 68000 and ECS.
By 1990 the 486s were on the market and Macs were shipping with faster 030s and could be equipped with NuBUS graphics cards that made Amiga graphics modes look like decelerated CGA.
After the A2000 the writing was on the wall.
Note: my perspective is of someone who has always used computers to do work, with ALMOST no care for video games so all of the blitter magic of Amiga was irrelevant to me. That being said when DOOM came out I bought a PC and rarely used my Amigas again.
What I can confidently assert is that I upgraded my A2000 many times and ran into the absolute configuration nightmare that is the Amiga architecture and the problems with grafting upgrades onto a complex system with multiple tiers of RAM and close OS integration with custom chips.
One more bit of heresy is that I always considered Sun's platform to be superior to SGI's.
> Amiga was only better 1985-1988. By 1987 PC and Mac caught up and never looked back.
Oh indubitably! I don't think even the most committed Amiga fan, even the ones that speculate about alternate histories, would deny that at all.
The thing is, though, that only happened because Commodore essentially decided that since it had so much of a head start, it could just rest on its laurels and not really innovate or improve anything substantially, instead of constantly pushing forward like all of its competitors would do, and so eventually the linear or even exponential curve of other hardware manufacturers' improvements outpaced its essentially flat improvement curve. So it doesn't seem like IBM PCs and eventually even Macs outpacing the power of Amiga Hardware was inevitable or inherent from the start.
If they had instead continued to push their lead — actually stuck with the advanced Amiga chips that they were working on before it was canceled and replaced with ECS for instance — I certainly see the possibility of them keeping up with other hardware, and eventually transitioning to 3D acceleration chips instead of 2D acceleration chips when that happened in the console world, eventually perhaps even leading to the Amiga line being the first workstation line to have the gpus, and further cementing their lead, while maintaining everything that made Amiga great.
Speculating even further, as we are seeing currently with the Apple M-series having a computer architecture that is composed of a ton of custom made special purpose chips is actually an extremely effective way of doing things; what if Amiga still existed in this day and age and had a head start in that direction, a platform with a history of being extremely open and well documented and extensible being the first to do this kind of architecture, instead of it being Apple?
Of course there may have been fundamental technical flaws with the Amiga approach that made it unable to keep up with other hardware even if Commodore had had the will; I have seen some decent arguments to that effect, namely that since it was using custom vendor-specific hardware instead of commodity hardware that was used by everyone, they couldn't take advantage of the cross-vendor compatibility like IBM PCs, could and also couldn't take advantage of economies of scale like Intel could, but who knows!
That's definitely how it seems to me, which is why I focused on Commodores poor management decisions first and only mentioned the possible technical issues second
The thing with Commodore was that as a company it was just totally dysfunctional. The basically did little useful development between C64 and the Amiga (the Amiga being mostly not their development). The Amiga didn't sell very well, specially in the US.
The company was going to shit after the Amiga launched, it took a competent manager to save the company and turn the Amiga around into a moderate success.
Commodore didn't really have money to keep up chip development. They had their fab they would have need to upgrade that as well, or drop it somehow.
Another example of that is the Acorn Archimedes. Absolutely fucking incredibly hardware for the price. Like crushing everything in price vs performance. But ... literally launched with a de-novo operating system with 0 applications. And its was a small company in Britain.
The dream scenario is for Sun to realize that they should build a low cost all costume chip device. They had the margin on the higher end business to support such a development for 2-3 generations and to get most software ported to it. They also had the software skill to make the hardware/software in a way that would allow future upgrades.
Imagining Sun buying Amiga and making it a lower end consumer workstation to pair with its higher end ones, with all the much-needed resources and interesting software that would have brought to the Amiga is a really cool thought experiment!
Sun did actually approach Commodore to license its technology for low end work station. However the Commodore CEO at the time declined for unknown reasons.
I don't know what Sun had planned for this tech.
A even more interesting approach for Sun would have been to cooperate or acquire Acorn. The Acorn Archimedes was an almost perfect low end work station product. Its incredible weakness was its lack of OS and it total lack of applications.
Acorn spend an absolutely absurd amount of money to try to get the OS and application on the platform. They spend 3 years developing an new OS, and then realized that this was going nowhere. So they rushed out another new OS. And then they realized that nobody want to buy a machine with a compromise OS and no application. So they had to put up huge effort to try to fix that. The company simply couldn't sustain that kind of effort on the Software side while at the same time building new processors and new machines. Its surprising what they achieved but it wasn't a good strategy.
Had they just adopted SunOS (BSD) it would have been infinitely better for them. And for Sun to release new high and and low end RISC workstations at the same time would have been an absolute bomb in the market.
Even if you added all the bells and whistles to the system (Ethernet, SCSI, extra RAM), you could be very low priced and absolutely blow pretty much every other system out of the water.
Re Acorn though — As much better from a market perspective as buying Acorn and releasing RISC- and BSD-based low-end workstations might have been for Sun, I still prefer to imagine a world where the Amiga's unique hardware and software got to live on — perhaps with compatibility layers to run Sun software, but nevertheless preserving a UNIX-like but still non-UNIX OS lineage and non-generic-PC hardware lineage.
It took a bit more than 1990, for PC 16 bit sound card, Super VGA screens, with Windows 3.1 to be widely adopted for the PC to out perform the Amiga, specially in European price points.
My first PC was acquired in 1992, and still only had a lousy beeper, on a 386SX.
I was similar, not really interested in graphics, just a nice programming environment. PCs had that stupid segmented address space (which was not ignorable at the programming language level), expensive tools, and crappy OSes. My Amiga 2000 had a flat address space, a nice C development environment, and multitasking actually worked. It really was ahead of its time, in combining a workstation-like environment and an affordable price.
Chip ram, fast ram, cpu ram, expansion board ram, or slow ram? Did too much ram force your zorro card into the slooooooooooow ram address space (mine did)? Tough cookies bucko!
As someone trying to get into Amiga retro competing as a hobby in today's day and age, I find it keeping all the different types of ram straight very confusing lol
It's the same in most computers. Wiping RAM is effort.
The feature here is that AmigaOS will try and reuse the ExecBase structure if found.
Such structure has a checksum, which is checked. If the check fails, a new one is made. This happens e.g. on power on, or after running games that are not system friendly (i.e. most games).
But if the check passes, this structure has important information, such as a list of memory regions, the "cold/cool/warm" vectors, which are function addresses that get called if non-zero at different points of the boot process (non-surprisingly a virus favorite), as well as and a list of reset-resident modules, which become allocated memory, thus protecting them.
A popular such device implements a reset-resident memory-backed block device, which the Amiga is able to boot from.
We kept our A2000 viable longer by adding the CPU board with the 030 chip. We went from 7MHz to somewhere around 40MHz or whatever. It meant that my Lightwave render went from 24 hours per frame to a few hours per frame.
I think you are mostly right, I just think your timing is off. Those early 386 machines and Mac II systems were very expensive, at least 2 to 3x the cost of an Amiga. The average home user wasn't going to drop $8K on a PS/2 model 80 with a 386/16.
By the early 90's the Amiga just wasn't competitive. The chip set barely evolved since 1985. ECS barely added anything over the original chip set. By around 1992 or 1993, 386 systems with SVGA and Soundblaster cards were cheap. Amiga AGA was too little, too late. Also consider the low end AGA system (Amiga 1200) was totally crippled with only 2 megs of slow "chip" RAM.
I was an Amiga fan until 1993. Started with an A500, then A3000. Eventually I moved on to a 486 clone w/Linux. Later on I had a Sun SparcStation 10 at home, so I agree with you on Sun and SGI.
The Amiga couldn't handle the performance requirements of Doom at the time (Game Engine Black Book Doom). Workbench was more fun than Windows and at least the install process that was early linux.
As much as I loved my O2 (my first professional computer), it was underpowered for the time for anything other than texture manipulation. The closed source nature of that time period and the hardware sales motion meant that you were paying through the teeth for compilers on top of already very expensive hardware. The Cray-linked Origin 200's ran Netscape web server with ease but that's a lot of hardware in a time period when everything went out of date very quickly-donated ours! Irix still looks better than the new Mac OS UIs IMO but no-Motif is a small price to pay for far cheaper access to SDKs IMO. Also, Irix was hilariously insecure due in part to its closed source nature. https://insecure.org/sploits_irix.html
"... hardware sales motion meant that you were paying through the teeth for compilers..."
For Fortran? My memory is hazy but at NASA NAS a bunch of us were using gcc/g++ starting ~1990. g++ was... an adventure. Building my own (fine!) compiler for free(!) got me hooked on OSS to the point that when Linux/FreeBSD launched I jumped in as fast as I could.
I really loved my various SGI boxen. Magical times. I was a NASA "manager" so had the Macintosh "manager" interface box that I solved by keeping it turned off.
We had 6 O200s Cray-linked into three nodes in an CXFS cluster to run an appletalk server backed by clarion arrays. While there were serious limits caused by the single metadata server, XVM and CXFS were better than anything provided by Veritas or the other major UNIX vendors of the day.
The Fibre Channel XIO boards were really needed back then for that application as PCI was still way too slow.
I was sad to know that when I left that job that SGI server was being replaced and the support personal at SGI were going to lose their jobs too.
The reason SGI failed, and eventually Sun too, isn't because the world "chose wrong", but because their performance simply did not keep up with x86.
When these RISC-based workstations were initially released their performance, especially at graphics, was well beyond what a PC could do, and justified their high prices. A "workstation" was in a class by itself, and helped establish the RISC mystique.
However, eventually Intel caught up with the performance, at a lower price, and that was pretty much the end. Sun lived on for a while based on their OS and software ecosystem, but eventually that was not enough especially with the advent of Linux, GCC, etc, as a free alternative.
Sun really struggled to make full use of their multicore systems. That m:n process model is coming back with fibers and libuv, but we have programming primitives and a deeper roster of experienced devs now than we did then. Back then they caused problems with scalability.
There were times when Java ran better on Intel than on Solaris.
Sun had the perfect opportunity with Utility Computing around the mid-2000s but when cloud took off we had Oracle buying SUNW. They killed Sun Cloud which had the opportunity to be big, vast, and powered by JAVA hardware.
Sun Microsystems was a company like no other. The last of a dying breed of "family" technology companies.
I was at the MySQL conference when it was announced that Oracle was buying Sun. It just took all the life out of the conference. All the Sun folks were super pissed off. Truly the end of an era.
I remember that time. It felt like Sun was on death's doorstep since the dot-com crash. On the hardware side, the market was flooded with used Sun hardware. On the software side, Linux was "good enough" for most workloads.
I was there too. It certainly felt "timed" to maximize the sense of deflation for people working on MySQL. Perhaps it was just coincidence. IIRC Larry Ellison said that the crown jewel in the deal was actually Java.
Ivan Sutherland described the reason [1] why PCs won a long time ago. Basically a custom tool may do a thing "better" than a general purpose tool for a while, but eventually, because more resources are spent improving the general tool, the generalized tool will be able to do the same thing as the specialty tool, but more flexibly and economically.
SGI dug their own grave. Not only were the workstations expensive, but they demanded outrageously priced support contracts. This behavior drives people nuts and will insure that the switch to a competitor the instant it becomes an option. Despite the high cost, the support contracts had a pretty lousy reputation as well, with long wait times for repairs from a handful of overworked techs. Even worse is the company turned away from its core competencies to focus on being an also-ran in the PC workstation market.
There was a window in the mid-90s where it would have been possible for SGI to develop a PC 3D accelerator for the consumer market using their GE technology, but nobody in the C-Suite had the stomach to make a new product that would undercut the enormous profit margins on their core product. It's the classic corporate trap. Missing out on the next big thing because you can't see past next quarter's numbers. Imagine basically an N64 on a PCI card for $150 in 1996. The launch versions could be bundled with a fully accelerated version of Quake. The market would have exploded.
I wish we could have a debugging view of the universe, draw a diagram with clusters of people labeled with company names, and watch them change over time. :-)
Maybe irix was okay to use if you were just sitting in front of it doing rando user / graphics things, but administering it was unbearable. The license fees to get OS updates were exorbitant; you'd have to get wacky new licenses to enable NFS or NIS and you'd need new kernels for just about anything.
As far as I could tell they were a cursed company that hated their users. "Here's a pretty thing that does one thing well but is otherwise insane and will ruin you when you need it most."
On HP-UX 10, back in 2000, the C compiler version I was using still wasn't fully ANSI C, and needed K&R C function declarations, but hey at least we had containers (HP Vault), and 64 bit file system access.
I used it for a while earlier at work, and don't remember many problems with it. One did have to apply OS patches fairly regularly to it, but IIRC, that process was somewhat smooth.
I never had an Amiga, but I had friends that had it. It was a superior tech only for a very small period of time.
What happened was Intel, they took great decisions like automating the design of their processors and this made them grow at an incredible pace. The Amiga depended on a different processor that stagnated.
The 68k CPU lineup at the heart of the Amiga was competitive well into the 90's; the Amiga had run out of juice by 1989. The Amiga was only as good as the custom chips. If Commodore kept investing in R&D for the custom chips, they would have at least remained competitive.
I used some SGIs in the mid-late nineties, and they did have cool 3D graphics capabilities. I found 4dwm to be kind of cool but mostly gimmicky and it was really slow on the Indy and O2. Windows 95/NT were much snappier on contemporary hardware.
By '97 or so SGI actually had essentially given up competing when they shut down the team that was developing the successor to InfiniteReality.
In a sense though, Silicon Graphics did become more standard, in that their original 3D framework was Iris GL, which then evolved into OpenGL, which became the main 3D graphics standard for many years.
I played with SGI machines in college and they felt like.. the future. I really hoped they would hire me when I graduated.
Incredible, though, how the relatively cheaper Windows NT machines and 3dfx cards and graphics software just killed them. I was a little sad when I wandered around the campus of an employer in Mountain View and noticed the fading sign that had what was left of the SGI logo.
The awesome old cube logo or the new "we spent millions of dollars on a professional marketing department to design a new logo" that is just the initials in a boring font and off center?
I co-oped for SGI onsite in the sales/marketing/support for a major ISP of the day back in the late 90s and the buzz around the office was that the company (at this point experimenting with overpriced Windows NT boxes and generic Linux servers) was experiencing massive brain drain to some brand new startup that was going to make something called a "GeForce" card for cheap PCs that was going to avoid the pitfalls of the then popular Voodoo cards. Apparently the engineers were unhappy with the direction the company was taking under the new leadership and thought that there was still an interest in graphics acceleration.
The "sgi" logo was a big step down from the cube, but it was a lot more attractive than the Rackable/Silicon Graphics International "sgi" logo that looked like a cheap knockoff of the previous one.
It really was a letdown when Rackable resurrected SGI and then brought about that ... thing of a logo. It just felt it hollowed out the brand even more, even if SGI itself was still making some interesting hardware at the time-namely the Altix 4700, UV and ICE, the soul just wasn't there anymore.
> I played with SGI machines in college and they felt like.. the future.
I had a couple of Indigos that I supported while an undergraduate (I had a student job with the University's Unix group in their computing center), and the SGIs felt to me exactly like the Amiga- Really cool, but kind of lopsided. I tended to do most of my work on the SPARCstations and ignore the SGIs unless I specifically wanted to play with the graphics stuff.
I actually still have an Indigo XS24 that I collected at one point over the years. Tried to get it to boot a bit ago but it's dead, unfortunately.
I was there near the end. First, as a summer intern in 1998, and then in 1999 as a full time engineer on what is now Google's Mountain View campus. SGI had always been a dream company for me. I'd first learned about them in high school; now, right out of college, I'd somehow managed to land a dream job.
SGI's hardware was cutting-edge and exotic. IRIX was killer (sorry Solaris). Cray was a subdivision. My coworkers used emacs, too. They put an O2 on my desk!
If a may, can I fact check a story conveyed to me through a mutual acquaintance of ours? The story was that SGI was trying to sell off MineSet, and needed the team to stick around long enough to sell them off -- so a bonus was to be given after a short period of time (a month maybe?). The bonus was significant enough to get people to at least defer a job search ($10K?), but SGI didn't manage to find a buyer. The check was to hit bank accounts on a particular day; the team waited to hear word that the literal money was in the bank -- and then all quit simultaneously.
Is there at least some truthiness to it? Or has this just become Silicon Valley urban legend in my head?
That rings a bell although fuzzily: as the new kid from school, I was pretty disconnected from the politics of the moment. I do seem to remember that the MineSet team departed en masse, but IIRC that departure roughly coincided with broader layoffs in the org.
(With apologies for reviving 90s IRIX/Solaris snark in my earlier post. :-)
I worked at Google from 2013 to 2020. There were definitely employees (maybe a majority) who assumed that Google would always be the dominant force in technology. Those of us who were a bit older always understood that everything changes in Silicon Valley.
Those buildings represented that change to me. I can remember coming to concerts at the Shoreline in the 90s and looking at those Silicon Graphics buildings: they looked so cool, and they represented the cutting edge of technology (at the time). And yet...it all disappeared.
Same goes for the Sun campus which is where Meta/Facebook is now. Famously, the Facebook entrance sign is literally the same old Sun sign, just turned around! [0]
So I always cautioned co-workers: this too, shall pass. Even Google.
I graduated undergrad in 1998 and can confirm that SGI was the company to go to. I felt so jealous of those few guys who had SGI offers, where I had to settle for a more generic PC graphics company. History is what it is but the SGI really had that luster that only a handful of companies ever boasted.
I had to support an open source library for all major unixes and the Irix compiler was by far the best one. It took years for the rest to catch up. But it took ages to compile with optimizations on. Good times.
It was a dream company for pretty much every siggraph person at that time. I was in grad school, eagerly awaiting a very popular 3-semester course in computer graphics. It had been devised and taught by a young promising professor who had published some pioneering siggraph papers. I signed up for the course. On the first day of class, the head of the department walked in and said the professor had been recruited by his dream company SGI for an ungodly sum of money to work on some Jewish director’s movie about a dinosaur themepark. I thought ok, whatever, someone else will teach the course. The bastards scrapped the entire 3 series computer graphics module because there wasn’t anyone else who could teach that. So we had to pick from one of the usual dumb options - databases, OS, Networks, Compilers. Since then I’ve always held a grudge against sgi.
> to work on some Jewish director’s movie about a dinosaur themepark
I assume you mean Steven Spielberg and one of the Jurassic Park films?
If so, why can't you just say so? Why are you referring to Steven Spielberg, one of the most famous directors of all time, as "some Jewish director?" Do you think people won't recognize the name? I promise people know who Steven Spielberg is.
I think the GP was telling their story in the context of that time. It's a technique to help the reader more fully understand the context. I'm almost sure there is a term for this literary technique.
This was their downfall, trying to scale out adoption with esoteric hardware.
I remember being quoted $18k ish for memory upgrade on a O2 or origin, same amount of memory I had just bought for $500 for an intel Linux box at home.
Sure, it wasn’t apples to apples, but I remember thinking very clearly that this wasn’t going to end well for SGI.
That's funny because my reaction to the O2 was "oh, this is far too expensive for what it is". Was workin on N64 game, and the other teams were using the Indy devkits, while we had PCs with the SN systems dev kits. Writing was on the wall at that point.
Yeah, the O2 definitely was too expensive for what it was. And while it was the least cool and powerful of the lineup by far, as a recent college grad, it was still the coolest computer I had ever had on my desk. ;-)
I was a system engineer for SGI in 1992 working mainly with McDonnell-Douglas in St Louis. It was thrilling to be sitting in the cafeteria and have Jim Clark plop down next to me for lunch. Just one of the guys.
As an outsider--cause I didn't live and work in California--this was the go-go atmosphere of such companies back then where they thought they could do no wrong. And the after work parties were wild (how the heck do you break off half a toilet bowl?).
One of the buildings had plastic over the windows cause that's where they were working on the plugin GL card for the PC. (Ssh! No one's supposed to know that!)
Being the first system engineer in St Louis, my eyes lit up when my manager told me he had ordered an 16-core machine for my office--just for me!
I was hired as a video expert. The company re-org'ed and my new boss decided he needed a Fortran expert so that was the end of my job with SGI.
When I was a kid I remember my brother and I asking my dad a ton of questions about SG. We viewed them as this amazing awesome company that made cool looking towers and the fastest computers in the world.
Much of this history is also described in Michael Lewis’s book the New New Thing (2000) which profiles Clark and his various ventures. It’s really a snapshot of pre-.com crash Silicon Valley.
I was in love with SGI when I was an undergraduate just over the hill at UC Santa Cruz in the early to mid 90s. Everything about the machines from their industrial designed cases but wonderfully colorful cases, and the sexy desktop OS ("This is UNIX. I know this!") and the way IrisGL rendered molecular graphics.
Driving to a Phish show at Shoreline, we passed the low-slung office buildings of SGI which seemed like the sexiest place to work. When I graduated, I thought I was "too dumb in CS" to get a job in Mountain View and went to grad school in biophysics instead.
By the time I was a few years into grad school, I worked in a computer graphics lab outfitted with Reality Monsters and Octanes and other high end SGIs (when you maxxed out an SGI's graphics and RAM, they were really fast). I was porting molecular graphics code to Linux using Mesa (much to the derision of the SGI fans in the lab). When we got a FireGL2 card it had a linux driver and could do reasonable molecular graphics in real time and the SGI folks looked real scared (especially because the SGI Visual Workstation had just come out and was a very expensive turkey).
Less than a decade after that I was working in those very buildings for Google. Google took over SGI's old HQ (Jeff Dean told me there was a period where Google and SGI overlapped in the GooglePlex and the SGI folks looked very sad as they paid for their lunches and teh googlers got free food). There was still plenty of SGI signage strewn about. And now Google has gone dumb and also built their own HQ next door (note the correlation between large SV companies building overly fancy HQs and then going out of business).
We've talked before about our respective molecular graphics background.
I started with Unix on a Personal IRIS as an undergrad working in a physics lab which used it for imaging capture and analysis. I was the nominal sys admin, with one semester of Minix under my belt and just enough to be dangerous. (I once removed /bin/cc because I thought it was possible to undelete, like on DOS. I had to ask around the meteorology department for a restore tape.)
The summer before grad school I got a job at the local supercomputing center to work on a parallelization of CHARMm, using PVM. I developed it on that PI, and on a NeXT. That's also when I learned about people at my future grad school working on VR for molecular visualization, in a 1992 CACM article. So when I started looking for an advisor, that's the lab I chose, and I became the junior co-author and eventual lead developer of VMD.
With a Crimson as my desktop machine, a lab full of SGIs and NeXTs, and the CAVE VR setup elsewhere in the building. Heady times.
I visited SGI in 1995 or so, on holiday, thinking that would be a great place to work. They even had an Inventor plugin for molecular visualization, so I thought it would be a good lead. I emailed and got an invited to visit, where the host kindly told me that they were not going to do more in molecular visualization because they wanted to provide the hardware everyone uses, and not compete in that software space.
In the early 1990s SGIs dominated molecular modeling (replacing Evans & Sutherland), so naturally the related tools, like molecular dynamics codes, also ran on them. But we started migrating to distributed computing, where it didn't make sense to have 16 expensive SGIs, leaving them more as the head .. which as you pointed out, was soon able to run just fine on a Linux machine.
Pardon my ignorance, but what is so unique or special about molecular visualisation compared to, say, Quake - or CAD? If you’ll permit me to reduce it down to “just” drawing organo-chem hexagons, lines, and red/grey/black spheres connecting those lines (and a 360-degrees spin animation for the investor-relations video) - where’s the room for the rest of CG? E.g., texture-mapping, fragment shaders, and displacement mapping?
Quakes came out in, what, 1996? And was written by some of the foremost practitioners of computer graphics?
We were a couple of physics grad students working on a side project in late 1993. My background was a semester course based on Foley & van Dam. Hardware gave us a 5-10 year lead over what we could have done with consumer tech.
There wasn't really a "rest of CG". Only the highest-end SGI machines at the time had hardware texture mapping - most did it in software (see https://en.wikipedia.org/wiki/Extreme_Graphics).
We aren't talking 2D organo-chem hexagons, but 3D spheres and cylinders. Back around 1995 I posted some benchmarks to Usenet about the different approaches I tried (including NURBS), but I can no longer find a copy of it.
The straight-forward way is to render the spheres as a bunch of triangles, so, what, 50 polygons per sphere? Times 100,000 spheres = 5 million polygons. That was large for the time, but doable. Plus, during movement we used a lower level of detail.
What was Quake's polygon count?
Oh, and we're displaying animated molecules, including interacting with a live physics simulation, so no pre-computed BSP either.
Rastering spheres quickly on a PC was also possible then, which was RasMol's forte, but it was flat compared to having a couple hardware-based point lights plus ambient lighting.
Back in 1993, I was in college and working for the extended education department running their all their computer infrastructure.
One day, someone wheeled this approx. 3x3 foot sized box to my door and asked me if I wanted it. It was a SGI Onyx with a giant monitor sitting on top, with a keyboard and mouse.
I plugged it in and it sounded like an airplane taking off. It immediately heated up my entire tiny office. It was the 4th Unix I had ever played with (Ultrix, NeXT and A/UX were previous ones). It had some cool games on it, but beyond that, at the time, I had no use for it because A/UX on my Quadra950, was so much more fun to play with.
I don't even think I ever opened it up to look at it. I don't know what I was thinking. lol.
After realizing it did not have much going for it, I ended up just turning it on when the office was cold and using it as a foot rest.
I agree that these machines and their OS were too proprietary and
over-engineered to weather the PC revolution. But when I think back to
my days using Suns and SG (Indigo) the memory feels like driving a
Rolls Royce or Daimler with leather seats and walnut panels.
> The article doesn't mention the reason for the fall: less good but cheaper competitors
The article has this: ''As Bob Bishop took the reigns of SGI, things looked dark. AMD announced their 64 bit architecture in October, PC graphics had made massive strides while remaining significantly less expensive than SGI’s offerings, NT was proving to be a solid and less expensive competitor to UNIX, Linux was eating away at traditional UNIX market segments, and Itanium still hadn’t launched.''
I can agree with almost all of that statement but I object to the ''NT was proving to be a solid and less expensive competitor to UNIX'' part as mostly false in any mixed OS environment over which I'd ever been admin.
Well, do remember the cost of a UNIX license at the time (unless you were using BSD). If you didn’t have thousands of dollars on hand, NT was a good choice.
I had an Indigo2 on my desk in college. I moved back and forth between that an a NeXT cube that a colleague had in their lab. The NeXT was nice but SLOW. The Indigo2 wasn't especially fast but it was nice and could do visual things that just weren't available as readily on our alternatives. We had SunOS and Solaris systems that were mostly used for network and engineering projects, and I was engaged in some visualization work. When the O2 was announced I was quite sure it would be the solution to our speed issues.. around the same time, another colleague was the first to install a beta of Win95, and it did seem awfully pretty.
I remember when we got the first Indys at the uni.
That was magic like.
People nearly got into physical fights to use one of them.
People came in at night to use them as well.
I wonder if the uni is so locked down now that students can sit in the
lab all night.
Being a bit pragmatic in getting my actual thesis done I discovered
that there was all of a sudden, a lot more resources available on
one of the (older) Sun servers.
It's pretty simple why these Unix vendors all died. Linux and Intel chips. Sure you could get a really nice Sun system at the time with all of the redundancy tech built in, which cost around $50k. Or you can go and get 4 or 5 Linux servers from Dell running RedHat which by the early 2000's were faster too.
They had me at ‘Skywriter Reality Engine’. I’d used vaxen at Uni in the 80s, then got into industry using Symbolics Lisp machines, then had several dark years programming on DOS boxes. The return to a truly innovative workstation blew my mind
378 comments
[ 2.2 ms ] story [ 104 ms ] threadI’ve always wanted to play with Irix . The UI looks very intuitive.
The boxes are hard to find, so I went for emulation but I couldn’t get it any further than a boot screen in MAME.
Later on, I was able to do some computational work on an Altix 3700 with 256 sockets and 512G of RAM spread over four full-height cabinets with the nest of NUMAlink cables at the back), at the time running SuSE linux and that was wild seeing the 256 sockets being printed out with a cat /proc/cpuinfo. Now the same capabilities are available in a 4U machine.
The corporate lineage story is also just as interesting as the hardware they made as well. Acquisition, spinoff, acquisition, rename, acqusition, shutter, now perhaps just a few books and binders and memories in the few remaining personnal at HPE are all that's left (via Cray, via Tera, via SGI, via Cray Research).
RIP SGI
I remember the colors as being very different, from the photos, though.
The Personal Iris was a deep purplish-brown, and the Indigo was ... indigo.
Jim Clark sounds like my kinda guy. I made a hash of my teenage years, and barely squeaked in, with a GED, myself. It has all worked out, OK, in the end, though.
When I was in high school we had a lab full of SGI machines. They also never got used. Hundreds of thousands of dollars of computing equipment, and probably that much again in software licenses (at the commercial rate), just sitting there doing nothing. It was heartbreaking.
On a happy note, the SGI bus (a semi-trailer full of SGI machines demoing their capabilities) came to school one time. As a teenage nerd, getting to play with a refrigerator-sized Onyx2 was a good day.
There were all kinds of toys, though. There was a dedicated classroom setup for video-based remote learning some 30 years before COVID - that got used for one semester, from what I gather (was never used while I was there). The school was even host to a dialup ISP at one point.
The administrators were all in on technology. The teachers, not so much...
Eventually, in my last year, the government changed the funding model and the party ended.
Nice.
I once worked at a startup that had a Cobalt Qube in the server room, and the Cobalt was ... cobalt blue.
https://en.m.wikipedia.org/wiki/File:Cobalt_Qube_3_Front.jpg
https://en.m.wikipedia.org/wiki/Cobalt_Qube
>Personal
>Iris
Must have been the color of their lover's eyes.
https://en.m.wikipedia.org/wiki/Iris_(given_name)
One of the neatest things is that they let us (Trilogy/pcOrder) put together a sand volleyball team to compete in their company intramurals.
Their cafeteria was also top notch.
https://en.m.wikipedia.org/wiki/The_Google_Story
Similarly, it feels like Silicon Graphics is a case where they really should have become more standard. Now, unlike Amiga, they were too expensive to catch on with regular consumers, but I feel like they should have become and stayed the "standard" for workstation computers.
Irix was a really cool OS, and 4Dwm was pretty nice to use and play with. It makes me sad that they beaten by Apple.
I think it's a combination of a skillset/culture needed to create a paradigm shift isn't the same one needed to compete with others on a playing field you built, and of complacency. It happens over an over. We saw it happen with RIM, and we're watching it happen right now with Prusa Research.
I think you highlighted very correctly there, though, why SGI lost. It turned out there were cheaper options, which while not on par with SGI workstations initially, just improved at a faster rate than SGI and eventually ended up with a much better cost/functionality profile. I feel like SGI just bet wrong. The article talks about how they acquired Cray, which were originally these awesome supercomputers. But it turned out supercomputers essentially got replaced by giant networks of much lower cost PCs.
I guess it's just kind of impossible to predict the future. I don't think it's an incompetent decision to try and focus entirely on the workstation world; there are lots of businesses that make no attempt to market to consumers, and only market to large companies/organizations, since the way budgeting works with big companies is sort of categorically different than consumer budgets.
But you're absolutely right. Apple and Windows computers just kept getting better and better, faster and faster, and cheaper and cheaper, as did 3D modeling and video editing software for them. I mean, hell, as a 12 year old kid in 2003, I had both Lightwave 3D (student license) and Screenblast Movie Studio (now Vegas) running on my cheap, low-spec desktop computer, and it was running fast enough to be useful (at least for standard definition).
I think a more useful explanation is that people rate the value of avoiding vendor lockin extraordinarily high, to the extent that people will happily pick worse technology if there's at least two competing vendors to choose from. The IBM PCs were not good, but for convoluted legal reasons related to screwups by IBM their tech became a competitive ecosystem. Bad for IBM, good for everyone else. Their competitors did not make that "mistake" and so became less preferred.
Microsoft won for a while despite being single vendor because the alternative was UNIX, which was at least sorta multi-vendor at the OS level, except that portability between UNIXen was ropey at best in the 90s and of course you traded software lockin for hardware lockin; not really an improvement. Combined with the much more expensive hardware, lack of gaming and terrible UI toolkits (of which Microsoft was the undisputed master in the 90s) and then later Linux, and that was goodbye to them.
Of course after a decade of the Windows monopoly everyone was looking for a way out and settled on abusing an interactive document format, as it was the nearest thing lying around that was a non-Microsoft specific way to display UI. And browsers were also a competitive ecosystem so a double win. HTML based UIs totally sucked for the end users, but .... multi-vendor is worth more than nice UI, so, it wins.
See also how Android wiped out every other mobile OS except iOS (nobody cares much about lockin for mobile apps, the value of them is just not high enough).
What smaller businesses are using will tend to be what takes over in the future, just due to natural processes. When smaller businesses grow, they would generally prefer to fund the concurrent growth of existing vendors that they like using than they are to switch to the existing "industrial-grade" vendor.
At the same time, larger organizations that can afford to start with the industrial-grade vendors are only as loyal as they are locked in.
But I think you're broadly right.
[1] Yes I know about OpenFOAM, I know I could use that if I really wanted.
So in 20 years in, the current batch of senior devs will be retiring, and the current noobies will have become senior devs.
*Whatever language is easy to learn today will be a big deal in 20 years*
That's how PHP, Python, and JavaScript won. Since JavaScript got so much money poured on it to make it fast, secure, easy, with a big ecosystem, I say JS (or at least TS) will still be a big deal in 20 years.
The latest batch of languages know this, and that's why there are no big minimal languages. Rust comes with a good package manager, unit tester, linter, self-updater, etc., because a language with friction for noobies will simply die off.
One might ask how we got stuck with the languages of script kiddies and custom animated mouse cursors for websites. There's no other way it could turn out, that's just how people learn languages.
My take: that’s just fine. Tightly crafted code is not a lost art, and is in fact getting easier to write these days. You’re just not forced into scrabbling for every last byte and cpu cycle anymore just to get acceptable results.
In 91 I was a dedicated Atari ST user convinced of the superiority of the 68k architecture, running a UUCP node off my hacked up ST. By the end of 92 I had a grey-box 486 running early releases of Linux and that was that. I used to fantasize over the photos and screenshots of workstations in the pages of UnixWorld and similar magazines... But then I could just dress my cheap 486 up to act like one and it was great.
When you switched to Intel in 1992, PC's had already existed since 1981. PC's didn't wipe out most other home computers overnight.
One day ...
I'm on board for this project?
Along the same lines, there is an alternate timeline where the Sharp X68000 took over the world: https://www.youtube.com/watch?v=OepeiBF5Jnk
Yeah, I think that would also have been a better timeline; I'm just stuck in the anglo-world and thus my knowledge is mostly limited to what was released in the US or Europe.
See the dominance of Threadripper in workstations, which is built on top of mainstream desktop and server parts bin. Or look at the Epyc based supercomputers, rumored to be the only supercomputers to turn a net profit for the suppliers, thanks to leveraging a lot of existing IP.
In this case, I think the market "chose right" - and the reason that the cheaper options won is because they were just better for the customer, better upgradability, better compatibility, and better competition among companies inside the ecosystems.
One of the most egregious things I point to when discussing SGI/Sun is how they were both so incredibly resistant to something as simple as the ATX/EATX standard for motherboard form factors. They just had to push their own form factors (which could vary widely from product to product) and allowed almost zero interoperability. This is just one small example but the attitude permeated both companies to the extent that it basically killed them.
How is that in anyway different from Apple today with it's ARM SoCs, soldered SSDs and an OS that requires "entitlements" from Apple to "unlock" features and develop on?
Interesting. How cheap? Never used Macs, only Windows and Unix and Linux.
There is no way to do this for an IBM z16, which is the kind of vendor lock in that people are saying Apple doesn’t have.
It hardly matters nowadays for most game developers.
In my opinion, Mesa played a more significant role because it first allowed people to port OpenGL software to run on software-only cheap systems running Linux, and later provided the framework for full OpenGL implementations coupled with hardware acceleration.
Of course, I still greatly enjoyed running Quake on Windows on my 3dfx card with OpenGL.
They leapt out ahead of the competition with an advanced OS, purpose-built for graphics and sound in a way that PCs and Macs weren't.
Which was great. But they weren't really better than the competition. They were just doing something the competition wasn't. And when the competition actually started doing those things they got eclipsed in a hurry.
I wonder if Tesla will suffer the same fate. They were obviously around a decade ahead of the established players when it came to electric cars. But once the other established players actually got serious about electric cars, Tesla largely stopped being special, and upstarts like Lucid and Rivian are neck and neck with them (in terms of compelling products, not sales) as well.
This means there are products out there with futuristic features that will be seen as requirements for all things going forward and right now those features are niche elements of some product.
The Amiga was a fantastic device but not a general purpose device. Lots of things are fantastic at a niche but not general, and those almost always fail.
Is this also the "worse is better" truism?
I don’t know anyone at Rivian so my opinion of them is neutral. Meanwhile Tesla is run by the jackass who ruined twitter.
I'm less clear on whether or not he's a net minus. I don't know if his detractors outnumber his fans or vice-versa.
At the time. A brief moment in time, and then they had no path forward and were rapidly steamrolled. Nothing was "chosen wrong" in this aspect.
There's other stuff too; they had better color graphics in the 80s while DOS was still dealing with CGA and EGA, and decent sound hardware. Even by 1990, the video toaster was released, well before it got any port to DOS.
[1] I'm sure it got better, my first exposure to it was System 7 and that thing was an unholy mess. I didn't touch macOS again until OS X.
> 80s while DOS was still dealing with CGA and EGA, and decent sound hardware.
And then the 80s ended. What point did I make that you are contradicting?
> Even by 1990, the video toaster was released,
And if you wanted to do CAD? Would you use an Amiga? Probably not. What about desktop publishing? Pointing out that Amiga had carved out a niche (in video editing) when that was the norm back in those days doesn't make any strong comment about the long term superiority or viability of the platform.
Also, I don't buy into the idea that just because a company had something "superior" for a short period of time with no further company direction that they didn't lose fair and square. That Amiga had something cool in the 80s but didn't or couldn't evolve isn't because the market "chose wrong". Commodore as a company was such a piece of shit it made Apple of the 80s look well run. Suffering a few more years with the occasional bomb on System 7 was not a market failure.
> Macs had bizarre cooperative multitasking
What was bizarre about it, compared to any other cooperative multitasking system of the time? Also you seem to be fixated on preemptive multitasking to the neglect of things like memory protection.
Yeah fair. I do wonder if a port like the SNES version would have been possible if id would have greenlit it, but that's a "what if" universe. Alien Breed 3D would run on a 1200, but IIRC it ran pretty poorly on that.
> And then the 80s ended. What point did I make that you are contradicting?
I mean, yes, VGA cards and Soundblaster cards were around in 1990, but they weren't really standard for several years later.
> And if you wanted to do CAD? Would you use an Amiga? Probably not. What about desktop publishing? Pointing out that Amiga had carved out a niche (in video editing) when that was the norm back in those days doesn't make any strong comment about the long term superiority or viability of the platform.
Also fair. I'll acknowledge my view is a bit myopic, since I don't really do CAD or desktop publishing, but I do some occasional video editing, and I do think Amigas were quite impressive on that front. You're right in saying it was a "niche" though.
> Commodore as a company was such a piece of shit it made Apple of the 80s look well run.
No argument here. Still think that the hardware was pretty cool though.
> What was bizarre about it
I guess "bizarre" was the wrong word. It was just really really unstable, and System 7 would constantly freeze for seemingly no reason and I hated it.
> Also you seem to be fixated on preemptive multitasking to the neglect of things like memory protection.
I feel like if Commodore had been competently run, they could have done work to get proper protected memory support, but again that's of course a "what if" universe that we can't really know for sure.
I guess what frustrates me is that it did genuinely feel like Commodore was really ahead of the curve. I think the fact that they had something pretty advanced like preemptive multitasking (edit: fixed typo) in the mid 80s was a solid core to build on, and I do kind of wish it had caught on and iterated. I see no reason why the Amiga couldn't have eventually gotten decent CAD and Desktop publishing software. I think Commodore didn't think they had to keep growing.
This is I think the premise that you and people like me who think Amiga could have gone on to do great things disagree on, I think. Most Amiga fans would say that it totally had a path forward, or at least there is no evidence that it didn't, and the failure to follow that path therefore it wasn't an inherent technical problem, but a problem of politics and management. Do you have any evidence to the contrary?
But the reality is the Commodore 64 kept Commodore going during most of that period rather than Amiga sales. It's similar to Apple where the Apple 2 kept Apple afloat during the 80s and 90s until Steve returned.
By the late 80s the "microcomputer" hobby/games market was dead and systems like the ST and Amiga (or Acorn Archimedes, etc.) were anachronisms. You had to be a PC-compat or a Mac or a Unix workstation or you were dead. Commodore and Atari both tried to push themselves into that workstation tier by selling cheaper 68030 machines than Sun, etc, but without success.
Amiga was only better 1985-1988.
I still have my original Amiga and A2000. I was an Amiga user for a decade. They were very good. I was platform agnostic, caring only to get work done as quickly and easily as possible so I was also an early Macintosh user as well as Sun and PA-RISC. And yes, I still have all of those dinosaurs too.
By 1987 PC and Mac caught up and never looked back.
But by 1988 the PS/2 with a 386 and VGA was out and the A2000 was shipping with a 7MHz 68000 and ECS.
By 1990 the 486s were on the market and Macs were shipping with faster 030s and could be equipped with NuBUS graphics cards that made Amiga graphics modes look like decelerated CGA.
After the A2000 the writing was on the wall.
Note: my perspective is of someone who has always used computers to do work, with ALMOST no care for video games so all of the blitter magic of Amiga was irrelevant to me. That being said when DOOM came out I bought a PC and rarely used my Amigas again.
What I can confidently assert is that I upgraded my A2000 many times and ran into the absolute configuration nightmare that is the Amiga architecture and the problems with grafting upgrades onto a complex system with multiple tiers of RAM and close OS integration with custom chips.
One more bit of heresy is that I always considered Sun's platform to be superior to SGI's.
Oh indubitably! I don't think even the most committed Amiga fan, even the ones that speculate about alternate histories, would deny that at all.
The thing is, though, that only happened because Commodore essentially decided that since it had so much of a head start, it could just rest on its laurels and not really innovate or improve anything substantially, instead of constantly pushing forward like all of its competitors would do, and so eventually the linear or even exponential curve of other hardware manufacturers' improvements outpaced its essentially flat improvement curve. So it doesn't seem like IBM PCs and eventually even Macs outpacing the power of Amiga Hardware was inevitable or inherent from the start.
If they had instead continued to push their lead — actually stuck with the advanced Amiga chips that they were working on before it was canceled and replaced with ECS for instance — I certainly see the possibility of them keeping up with other hardware, and eventually transitioning to 3D acceleration chips instead of 2D acceleration chips when that happened in the console world, eventually perhaps even leading to the Amiga line being the first workstation line to have the gpus, and further cementing their lead, while maintaining everything that made Amiga great.
Speculating even further, as we are seeing currently with the Apple M-series having a computer architecture that is composed of a ton of custom made special purpose chips is actually an extremely effective way of doing things; what if Amiga still existed in this day and age and had a head start in that direction, a platform with a history of being extremely open and well documented and extensible being the first to do this kind of architecture, instead of it being Apple?
Of course there may have been fundamental technical flaws with the Amiga approach that made it unable to keep up with other hardware even if Commodore had had the will; I have seen some decent arguments to that effect, namely that since it was using custom vendor-specific hardware instead of commodity hardware that was used by everyone, they couldn't take advantage of the cross-vendor compatibility like IBM PCs, could and also couldn't take advantage of economies of scale like Intel could, but who knows!
The company was going to shit after the Amiga launched, it took a competent manager to save the company and turn the Amiga around into a moderate success.
Commodore didn't really have money to keep up chip development. They had their fab they would have need to upgrade that as well, or drop it somehow.
Another example of that is the Acorn Archimedes. Absolutely fucking incredibly hardware for the price. Like crushing everything in price vs performance. But ... literally launched with a de-novo operating system with 0 applications. And its was a small company in Britain.
The dream scenario is for Sun to realize that they should build a low cost all costume chip device. They had the margin on the higher end business to support such a development for 2-3 generations and to get most software ported to it. They also had the software skill to make the hardware/software in a way that would allow future upgrades.
I don't know what Sun had planned for this tech.
A even more interesting approach for Sun would have been to cooperate or acquire Acorn. The Acorn Archimedes was an almost perfect low end work station product. Its incredible weakness was its lack of OS and it total lack of applications.
Acorn spend an absolutely absurd amount of money to try to get the OS and application on the platform. They spend 3 years developing an new OS, and then realized that this was going nowhere. So they rushed out another new OS. And then they realized that nobody want to buy a machine with a compromise OS and no application. So they had to put up huge effort to try to fix that. The company simply couldn't sustain that kind of effort on the Software side while at the same time building new processors and new machines. Its surprising what they achieved but it wasn't a good strategy.
Had they just adopted SunOS (BSD) it would have been infinitely better for them. And for Sun to release new high and and low end RISC workstations at the same time would have been an absolute bomb in the market.
Even if you added all the bells and whistles to the system (Ethernet, SCSI, extra RAM), you could be very low priced and absolutely blow pretty much every other system out of the water.
Re Acorn though — As much better from a market perspective as buying Acorn and releasing RISC- and BSD-based low-end workstations might have been for Sun, I still prefer to imagine a world where the Amiga's unique hardware and software got to live on — perhaps with compatibility layers to run Sun software, but nevertheless preserving a UNIX-like but still non-UNIX OS lineage and non-generic-PC hardware lineage.
My first PC was acquired in 1992, and still only had a lousy beeper, on a 386SX.
Chip ram, fast ram, cpu ram, expansion board ram, or slow ram? Did too much ram force your zorro card into the slooooooooooow ram address space (mine did)? Tough cookies bucko!
Macintosh, pounding on table: "RAM is RAM!"
The feature here is that AmigaOS will try and reuse the ExecBase structure if found.
Such structure has a checksum, which is checked. If the check fails, a new one is made. This happens e.g. on power on, or after running games that are not system friendly (i.e. most games).
But if the check passes, this structure has important information, such as a list of memory regions, the "cold/cool/warm" vectors, which are function addresses that get called if non-zero at different points of the boot process (non-surprisingly a virus favorite), as well as and a list of reset-resident modules, which become allocated memory, thus protecting them.
A popular such device implements a reset-resident memory-backed block device, which the Amiga is able to boot from.
By the early 90's the Amiga just wasn't competitive. The chip set barely evolved since 1985. ECS barely added anything over the original chip set. By around 1992 or 1993, 386 systems with SVGA and Soundblaster cards were cheap. Amiga AGA was too little, too late. Also consider the low end AGA system (Amiga 1200) was totally crippled with only 2 megs of slow "chip" RAM.
I was an Amiga fan until 1993. Started with an A500, then A3000. Eventually I moved on to a 486 clone w/Linux. Later on I had a Sun SparcStation 10 at home, so I agree with you on Sun and SGI.
As much as I loved my O2 (my first professional computer), it was underpowered for the time for anything other than texture manipulation. The closed source nature of that time period and the hardware sales motion meant that you were paying through the teeth for compilers on top of already very expensive hardware. The Cray-linked Origin 200's ran Netscape web server with ease but that's a lot of hardware in a time period when everything went out of date very quickly-donated ours! Irix still looks better than the new Mac OS UIs IMO but no-Motif is a small price to pay for far cheaper access to SDKs IMO. Also, Irix was hilariously insecure due in part to its closed source nature. https://insecure.org/sploits_irix.html
For Fortran? My memory is hazy but at NASA NAS a bunch of us were using gcc/g++ starting ~1990. g++ was... an adventure. Building my own (fine!) compiler for free(!) got me hooked on OSS to the point that when Linux/FreeBSD launched I jumped in as fast as I could.
I really loved my various SGI boxen. Magical times. I was a NASA "manager" so had the Macintosh "manager" interface box that I solved by keeping it turned off.
That was in addition to having three default accounts with well known passwords and a telnet server.
Oracle threatened to not support us when I used an unprivileged Xvfb instqance instead.
Still stupid but not that uncommon back then.
The Fibre Channel XIO boards were really needed back then for that application as PCI was still way too slow.
I was sad to know that when I left that job that SGI server was being replaced and the support personal at SGI were going to lose their jobs too.
When these RISC-based workstations were initially released their performance, especially at graphics, was well beyond what a PC could do, and justified their high prices. A "workstation" was in a class by itself, and helped establish the RISC mystique.
However, eventually Intel caught up with the performance, at a lower price, and that was pretty much the end. Sun lived on for a while based on their OS and software ecosystem, but eventually that was not enough especially with the advent of Linux, GCC, etc, as a free alternative.
There were times when Java ran better on Intel than on Solaris.
Sun Microsystems was a company like no other. The last of a dying breed of "family" technology companies.
[1] http://www.cap-lore.com/Hardware/Wheel.html
There was a window in the mid-90s where it would have been possible for SGI to develop a PC 3D accelerator for the consumer market using their GE technology, but nobody in the C-Suite had the stomach to make a new product that would undercut the enormous profit margins on their core product. It's the classic corporate trap. Missing out on the next big thing because you can't see past next quarter's numbers. Imagine basically an N64 on a PCI card for $150 in 1996. The launch versions could be bundled with a fully accelerated version of Quake. The market would have exploded.
Absolutely, they could have been where Nvidia is now!
Worked at a university in the early 90s.
Maybe irix was okay to use if you were just sitting in front of it doing rando user / graphics things, but administering it was unbearable. The license fees to get OS updates were exorbitant; you'd have to get wacky new licenses to enable NFS or NIS and you'd need new kernels for just about anything.
As far as I could tell they were a cursed company that hated their users. "Here's a pretty thing that does one thing well but is otherwise insane and will ruin you when you need it most."
Good riddance.
The Saturn 5 was clearly a technical marvel better than any plane, and it'd get you anywhere much faster.
If you spare no expense, you get a better product. Sure. I'm also not surprised that a $100k BMW is more comfortable than a Renault Clio.
I used it for a while earlier at work, and don't remember many problems with it. One did have to apply OS patches fairly regularly to it, but IIRC, that process was somewhat smooth.
And if memory serves, the Bible (https://www.goodreads.com/book/show/603263.Advanced_Programm...) didn’t cover it, which was a problem.
Regarding Windows, some time reading the excellent Windows Internals book series is recommended.
What happened was Intel, they took great decisions like automating the design of their processors and this made them grow at an incredible pace. The Amiga depended on a different processor that stagnated.
By '97 or so SGI actually had essentially given up competing when they shut down the team that was developing the successor to InfiniteReality.
In a sense though, Silicon Graphics did become more standard, in that their original 3D framework was Iris GL, which then evolved into OpenGL, which became the main 3D graphics standard for many years.
Incredible, though, how the relatively cheaper Windows NT machines and 3dfx cards and graphics software just killed them. I was a little sad when I wandered around the campus of an employer in Mountain View and noticed the fading sign that had what was left of the SGI logo.
I co-oped for SGI onsite in the sales/marketing/support for a major ISP of the day back in the late 90s and the buzz around the office was that the company (at this point experimenting with overpriced Windows NT boxes and generic Linux servers) was experiencing massive brain drain to some brand new startup that was going to make something called a "GeForce" card for cheap PCs that was going to avoid the pitfalls of the then popular Voodoo cards. Apparently the engineers were unhappy with the direction the company was taking under the new leadership and thought that there was still an interest in graphics acceleration.
https://en.wikipedia.org/wiki/Silicon_Graphics_International
I had a couple of Indigos that I supported while an undergraduate (I had a student job with the University's Unix group in their computing center), and the SGIs felt to me exactly like the Amiga- Really cool, but kind of lopsided. I tended to do most of my work on the SPARCstations and ignore the SGIs unless I specifically wanted to play with the graphics stuff.
I actually still have an Indigo XS24 that I collected at one point over the years. Tried to get it to boot a bit ago but it's dead, unfortunately.
3Dfx grew up in the arcade market. They were always consumer-focused.
SGI's hardware was cutting-edge and exotic. IRIX was killer (sorry Solaris). Cray was a subdivision. My coworkers used emacs, too. They put an O2 on my desk!
The dream didn't last long. Major layoffs hit just a few months after I started full time. I wrote about the experience here: https://davepeck.org/2009/02/11/the-luckiest-bad-luck/
Is there at least some truthiness to it? Or has this just become Silicon Valley urban legend in my head?
(With apologies for reviving 90s IRIX/Solaris snark in my earlier post. :-)
Those buildings represented that change to me. I can remember coming to concerts at the Shoreline in the 90s and looking at those Silicon Graphics buildings: they looked so cool, and they represented the cutting edge of technology (at the time). And yet...it all disappeared.
Same goes for the Sun campus which is where Meta/Facebook is now. Famously, the Facebook entrance sign is literally the same old Sun sign, just turned around! [0]
So I always cautioned co-workers: this too, shall pass. Even Google.
[0] https://www.businessinsider.com/why-suns-logo-is-on-the-back...
Great memento mori.
Which is one of the all-time greats IMHO. I'd keep it around too.
It was a dream company for pretty much every siggraph person at that time. I was in grad school, eagerly awaiting a very popular 3-semester course in computer graphics. It had been devised and taught by a young promising professor who had published some pioneering siggraph papers. I signed up for the course. On the first day of class, the head of the department walked in and said the professor had been recruited by his dream company SGI for an ungodly sum of money to work on some Jewish director’s movie about a dinosaur themepark. I thought ok, whatever, someone else will teach the course. The bastards scrapped the entire 3 series computer graphics module because there wasn’t anyone else who could teach that. So we had to pick from one of the usual dumb options - databases, OS, Networks, Compilers. Since then I’ve always held a grudge against sgi.
I assume you mean Steven Spielberg and one of the Jurassic Park films?
If so, why can't you just say so? Why are you referring to Steven Spielberg, one of the most famous directors of all time, as "some Jewish director?" Do you think people won't recognize the name? I promise people know who Steven Spielberg is.
Based on the comment, it sounds like that's the way the head of the department phrased it.
Presumably the department head didn't know the title as it hadn't been released yet.
This was their downfall, trying to scale out adoption with esoteric hardware.
I remember being quoted $18k ish for memory upgrade on a O2 or origin, same amount of memory I had just bought for $500 for an intel Linux box at home.
Sure, it wasn’t apples to apples, but I remember thinking very clearly that this wasn’t going to end well for SGI.
As an outsider--cause I didn't live and work in California--this was the go-go atmosphere of such companies back then where they thought they could do no wrong. And the after work parties were wild (how the heck do you break off half a toilet bowl?).
One of the buildings had plastic over the windows cause that's where they were working on the plugin GL card for the PC. (Ssh! No one's supposed to know that!)
Being the first system engineer in St Louis, my eyes lit up when my manager told me he had ordered an 16-core machine for my office--just for me!
I was hired as a video expert. The company re-org'ed and my new boss decided he needed a Fortran expert so that was the end of my job with SGI.
https://vizworld.com/2009/04/what-led-to-the-fall-of-sgi-cha...
https://vizworld.com/2009/04/what-led-to-the-fall-of-sgi-cha...
https://vizworld.com/2009/04/what-led-to-the-fall-of-sgi-cha...
https://vizworld.com/2009/04/what-led-to-the-fall-of-sgi-cha...
https://vizworld.com/2009/05/what-led-to-the-fall-of-sgi-epi...
Driving to a Phish show at Shoreline, we passed the low-slung office buildings of SGI which seemed like the sexiest place to work. When I graduated, I thought I was "too dumb in CS" to get a job in Mountain View and went to grad school in biophysics instead.
By the time I was a few years into grad school, I worked in a computer graphics lab outfitted with Reality Monsters and Octanes and other high end SGIs (when you maxxed out an SGI's graphics and RAM, they were really fast). I was porting molecular graphics code to Linux using Mesa (much to the derision of the SGI fans in the lab). When we got a FireGL2 card it had a linux driver and could do reasonable molecular graphics in real time and the SGI folks looked real scared (especially because the SGI Visual Workstation had just come out and was a very expensive turkey).
Less than a decade after that I was working in those very buildings for Google. Google took over SGI's old HQ (Jeff Dean told me there was a period where Google and SGI overlapped in the GooglePlex and the SGI folks looked very sad as they paid for their lunches and teh googlers got free food). There was still plenty of SGI signage strewn about. And now Google has gone dumb and also built their own HQ next door (note the correlation between large SV companies building overly fancy HQs and then going out of business).
Such is the cycle of sexy tech.
I started with Unix on a Personal IRIS as an undergrad working in a physics lab which used it for imaging capture and analysis. I was the nominal sys admin, with one semester of Minix under my belt and just enough to be dangerous. (I once removed /bin/cc because I thought it was possible to undelete, like on DOS. I had to ask around the meteorology department for a restore tape.)
The summer before grad school I got a job at the local supercomputing center to work on a parallelization of CHARMm, using PVM. I developed it on that PI, and on a NeXT. That's also when I learned about people at my future grad school working on VR for molecular visualization, in a 1992 CACM article. So when I started looking for an advisor, that's the lab I chose, and I became the junior co-author and eventual lead developer of VMD.
With a Crimson as my desktop machine, a lab full of SGIs and NeXTs, and the CAVE VR setup elsewhere in the building. Heady times.
I visited SGI in 1995 or so, on holiday, thinking that would be a great place to work. They even had an Inventor plugin for molecular visualization, so I thought it would be a good lead. I emailed and got an invited to visit, where the host kindly told me that they were not going to do more in molecular visualization because they wanted to provide the hardware everyone uses, and not compete in that software space.
In the early 1990s SGIs dominated molecular modeling (replacing Evans & Sutherland), so naturally the related tools, like molecular dynamics codes, also ran on them. But we started migrating to distributed computing, where it didn't make sense to have 16 expensive SGIs, leaving them more as the head .. which as you pointed out, was soon able to run just fine on a Linux machine.
Not looking it up right now but the original Q1 had a very low poly count.
We were a couple of physics grad students working on a side project in late 1993. My background was a semester course based on Foley & van Dam. Hardware gave us a 5-10 year lead over what we could have done with consumer tech.
There wasn't really a "rest of CG". Only the highest-end SGI machines at the time had hardware texture mapping - most did it in software (see https://en.wikipedia.org/wiki/Extreme_Graphics).
We aren't talking 2D organo-chem hexagons, but 3D spheres and cylinders. Back around 1995 I posted some benchmarks to Usenet about the different approaches I tried (including NURBS), but I can no longer find a copy of it.
The straight-forward way is to render the spheres as a bunch of triangles, so, what, 50 polygons per sphere? Times 100,000 spheres = 5 million polygons. That was large for the time, but doable. Plus, during movement we used a lower level of detail.
What was Quake's polygon count?
Oh, and we're displaying animated molecules, including interacting with a live physics simulation, so no pre-computed BSP either.
Rastering spheres quickly on a PC was also possible then, which was RasMol's forte, but it was flat compared to having a couple hardware-based point lights plus ambient lighting.
Interestingly, AutoCAD (RIP Walker) tried to get into molecular modeling, but it didn't work out. https://www.fourmilab.ch/autofile/e5/chapter2_82.html
One day, someone wheeled this approx. 3x3 foot sized box to my door and asked me if I wanted it. It was a SGI Onyx with a giant monitor sitting on top, with a keyboard and mouse.
I plugged it in and it sounded like an airplane taking off. It immediately heated up my entire tiny office. It was the 4th Unix I had ever played with (Ultrix, NeXT and A/UX were previous ones). It had some cool games on it, but beyond that, at the time, I had no use for it because A/UX on my Quadra950, was so much more fun to play with.
I don't even think I ever opened it up to look at it. I don't know what I was thinking. lol.
After realizing it did not have much going for it, I ended up just turning it on when the office was cold and using it as a foot rest.
Oh yea, found a video... https://www.youtube.com/watch?v=Bo3lUw9GUJA
The article has this: ''As Bob Bishop took the reigns of SGI, things looked dark. AMD announced their 64 bit architecture in October, PC graphics had made massive strides while remaining significantly less expensive than SGI’s offerings, NT was proving to be a solid and less expensive competitor to UNIX, Linux was eating away at traditional UNIX market segments, and Itanium still hadn’t launched.''
I can agree with almost all of that statement but I object to the ''NT was proving to be a solid and less expensive competitor to UNIX'' part as mostly false in any mixed OS environment over which I'd ever been admin.
I wonder if the uni is so locked down now that students can sit in the lab all night.
Being a bit pragmatic in getting my actual thesis done I discovered that there was all of a sudden, a lot more resources available on one of the (older) Sun servers.
It saved me days if not weeks.