In the latest contest, not only did Linux dominate, but Linux showed that is slowly pushing out all its competitors. In the June 2014 Top 500 supercomputer list, the top open-source operating system set a new high with 485 systems out of the fastest 500 running Linux. In other words 97 percent of the fastest computers in the world are based on Linux.
With numbers like this, it’s easy to forget that this project started with the words “just a hobby, won’t be big and professional like gnu”.
This hobby now dominates almost every field of computing – from mobile to supercomputing.
I think it must have been by 2005 that Linux had first been used for 50+% of the top 500 supercomputers. I remember it was 2009 that 92% of the top 500 were running Linux.
In hindsight it seems like a no brainer:
1. Most of the HPC people come from the Unix world.
2. Linux is like Unix and it’s POSIX compliant.
3. Linux cost is $0 *if* you’re an expert at OS administration which is true for… the HPC people who knew Unix.
4. Linux has had multicore support since 2.0 series from the mid 90’s, and GPU computing support since NVidia got into the GPU computing sector in the late 2000s.
5. None of that per CPU/GPU software licensing hassles means you can spend your money on hardware and support staff instead.
As I understand it, Linux isn’t actually POSIX compliant, but it’s close enough so that it does not matter for 99% of tasks.
But actually I think if you were making a top 100 supercomputer, Microsoft would gladly give you a license to Windows, or probably pay you to take it for the publicity. However, on supercomputers, the OS is a distraction from the task in hand, and all they want is an OS that works well enough and gets out of the way. We’re also talking about computer scientists who are likely good enough to tinker with the OS itself to tune it for their own needs, and Windows does not really suit that.
Linux is a good choice, of course, but I think the reason it’s used so often in embedded systems or supercomputers is much the same reason as Windows is used so often on the desktop, it’s got inertia, a code base, people know it.
It is complient, but nobody is paying the fee to have each release certified.
Of course POSIX is many standards, there are a few of them Linux doesn’t implement, or implement fully, but still more than most.
Huh. So when something is advertised as ‘POSIX compliant’, that could refer to any of many POSIX standards?
No actually there are some corner cases where Linux does the fast thing instead of the POSIX thing, such as select() very rarely returning read-ready for a socket that has nothing to read.
This happened due to arcane magic in the socket code so it may have been fixed and I didn’t notice.
I’m not an ideological open source cheerleader but let’s face it, it is simply fantastic and Linux is its standard bearer.
It fails occasionally; OS X.1 did what linux failed to do for the mainstream desktop experience after years and years of trying and professional media suits are sadly lacking. But, Linux/Gnu along with many others continue to exhibit why this is an incredibly valuable human endeavour.
I’ve had the pleasure of tinkering with Linux Yellow Dog a long time ago and still do so on a regular basis with other flavours on a myriad of devices.
Well, sort of. I’ll grant you it’s done a better job than Linux desktops, but OS-X is still a tiny small proportion of overall desktop use (~7%). Is that really enough to be considered “mainstream”, rather than “niche”?
Of course, both Linux (in the form of Android) and OS-X (in the form of iOS) have been *much* more successful in mobile…
As this post is really about the growth of a hobby project to an amazing thing that has added real value to pretty much everything, I shall resist my Apple zealotry instincts and just say: fair enough.
It’s just an unfortunate feature of the industry – it really doesn’t matter how good you are, it’s almost impossible to unseat an incumbent player with the weight of numbers on their side.
In mobile, that’s actually working in Linux’s (But not GNU’s) favor for once.
But they also didn’t need to get existing users to switch (unless you count feature phones). Smart phons was an exploding market when they got in, and iOS and Android got a big share of the new market, and only later the old market to switch when they already dominated.
The car industry is a classic example. Mercedes engineering prowess and quality has plummeted over the past 20 years. Jaguar now make very sophisticated and reliable cars. The Koreans now make the most reliable cars of all. Yet the message hasn’t really got through to the public whose opinions are extremely outdated.
Edited 2014-06-25 09:19 UTC
Mercedes are still okay, Audis are a better example. They have been the most unreliable car manufacturer since the early 2000s (according to UK ensurance companies), but they are still considered a high-end quality car because they were good in the 80s and 90s.
Linux seems to have most of the ‘bread and butter’ apps covered pretty well, but lacks in the professional/high-end desktop app department. I’ve heard that Macs have the exact opposite problem, although I’m not entirely certain about this. I’ve considered buying one for a home/hobbyist music studio.
Actually you’ll find that there are a ton of little “one app for a specific need” apps for the Mac, and most of them are quite polished. There’s pretty much nothing that I can’t do with the Mac vs my Crunchbang and Windows desktops, when it comes to productivity. I just prefer the workflow in Crunchbang to the other two.
For music creation you really can’t beat a Mac, even the built in GarageBand software is better than a lot of the paid Windows equivalents. And the price of Logic has come down a lot lately. Pro Tools works on Mac too.
I hear you about the workflow. Mac OS X seriously drives me nuts. When I first started using it I was thinking “Oh, did Gnome-shell grab this from OSX or did OSX grab it from Gnome-shell….
Then realized that Gnome-Shell is way better. Hopefully Gnome never does what OSX, iOS and Android do… when I close an application, I want it closed, I don’t want it sitting there in the task manager…
Use cmd+q then xD
Gnome-shell is “way better” for gnome developers and hardcore gnome fans… not for normal people and not even for advanced linux users like myself.
IMHO Gnome3 and KDE4 were the beginning of the end for major Linux Desktops. They just suck, suck for normal people and suck for nerds. Nobody wants them.
That’s why small projects like MATE or LXDE are more popular than ever. It’s very sad for me because I was a huge KDE3 fan… but that’s the truth. When you want to reinvent the wheel (badly) you got KDE4 and Gnome3.
Ha ha, Command+Q is what I have been using, but still, am confused on what exactly the difference between X and _ is… and they can’t even get maximize correct, so they remove it in 10.10?
I am a hardcore Linux user. Still have my Debian 1.3.1 CD.
And I actually find Gnome-Shell quite awesome and usable. Then again some of the features they randomly like to pull out and then wait 5 versions and then maybe put back in is kind of frustrating, but then they did that with gnome 1-3.
One thing I wish Apple would nick from Gnome is the ability to close an application through ‘Mission Control’ or whatever OSX calls it.
It’s honestly not that bad on the Mac, at least for me. But it’s definitely not the best either. I used nothing but Mac OS from 2005 to 2007 on a G4 mini because I got frustrated with the state of Windows and GNU/Linux at the time. When I got rid of the mini and went back to Ubuntu, I realized that Canonical had gotten a lot of stuff right in the interim.
Fast forward to today and I’ve scaled back to the minimalism of Crunchbang for daily use at home, and it’s even on one of my work computers. It has a ton of functionality built-in under the minimalist interface, if you choose to use it. It’s really a top-notch distro. When I switch over to my Intel Mac mini to work on music, it’s almost jarring. I don’t miss using OS X daily, but it gets the job done when it comes to media production. Maybe the newest versions are better at daily work, but I’m sticking with Snow Leopard to avoid another hardware investment.
I am still debating getting a PPC Mac… just so I can put MorphOS on it
It’s cheaper than trying to find a PPC accelerator for my A4000D.
If you get a G4 “Quicksilver” you can heavily mod is with a dual 1.8Ghz 7448 CPU upgrade and vBIOS flash a PC Geforce 7800GS or Radeon x800XL(?) AGP cars for something that’ll be faster then the early Intel iMacs as well as have the option for SATA drives.
Only problem is you’ll be limited to PC133 SDRAM and 4xAGP. On the other hand the MDDs are more recent Mac towers that have DDR ram, but lack CPU upgrade abilities and use a very strange PSU that can be a pain to source where as the Quicksilver, while using a non standard ATX pinout you can pick up an adapter for a standard ATX PSU and use it in them.
Amen about GarageBand. And iMovie. There is plenty of music software for desktop Linux, but the best is quite cumbersome. And there is nothing remotely comparable to GarageBand or iMovie for iOS in the Android world.
I have an Android phone and tablet, and while I am quite satisfied, i wish Google, or whoever else, produced something that could approach the GarageBand I had in my old iPhone 4. I doubt it is coming soon, unfortunately; Google seems uninterested on the “creativity” facet of mobile computing, aside of shoving pics and vids into the social networks.
“bread and butter” wot??
have you ever had a linux server?
You can install almost anything.. heh, really, “bread and butter” would fit a lot better on windows servers.
I can even install a microsoft domain controller on a linux server using samba4.
Linux not far from beating microsoft in their own game lol.
the internet is run by linux/BSDs
I was talking about desktops, not servers.
It seems to me that real time computing is the only remaining territory Linux has not made significant headway into so far.
“real time”
If You exclude large scale real time that is (like stock exchanges).
“Real-time” covers a lot of territory. Do you mean hard real-time e.g., flight control and nuclear power plants? Telecom? Financial systems? Games?
As far as I know, Linux covers all of the above bases very well, though it hasn’t dominated all of them quite yet. Both Red Hat and SUSE, for example, offer hard real-time products for their Linux environments, and they work quite well for flight simulations and such at major corporations. Dozens of stock exchanges run on these products as well, including such prime markets as New York and London. And of course Steam famously targeted Linux as a premium gaming platform a couple of years back. It’s only telecom where I haven’t seen Linux really take off yet – but then, I don’t work in telecom, so I may just be out of touch.
Funny how well a truly open and collaborative kernel can adapt to so many different environments. I suspect few people expected this back in the 90’s, before IBM’s billion dollar investment drew attention to the upstart OS with the non-proprietary license.
ricegf,
The desktop has long been controlled by network effects; I have to use/support windows because my clients use windows, etc. However in industries where network effects were less influential (ie many embedded devices), that’s where linux has been able to grow the fastest.
For massively parallel research labs writing their own software, they have no binary application lock in. Code with portable dependencies, can be recompiled for almost any platform, so they can choose their platform on merits.
> so they can choose their platform on merits.
Uf, we hope it’s that way, and that they don’t choose their platform on usual vendors techniques like getting paid disguised dinners, travels and hotels to “attend” “conferences” and “courses”, disguised presents, etc. 🙁
Edited 2014-06-25 17:00 UTC
Nth_Man,
Are you implying that linux bribed it’s way into the top 500? My gut reaction is to laugh out loud. It seems more likely that it was indeed a choice rather than coercion. The cost is good, the platform robust, the software repos are full of powerful tools, the hardware support is good. Windows can work too, but lets be honest MS has not been pushing the right levers to appeal to researchers and power users, neither of whom like being kept on a short chain tagging along behind microsoft’s whims. Freedom matters.
Edited 2014-06-25 17:44 UTC
> Are you implying that linux bribed it’s way into the top 500?
Linux didn’t do that, like you already knew. I was talking about some vendor tactics. If you read my comment correctly, you can see that those methods are used by the richest companies to force changes, to stop them, etc.
> My gut reaction is to laugh out loud.
Of course, some people get a text backwards and then they laugh. It says a lot. Naturally, don’t expect me to read or write any more comment in a thread displaying that level of education and understanding.
Edited 2014-06-25 20:38 UTC
Nth Man,
Even in hindsight, it’s still unclear from your post, which is exactly why my posted started with a question.
“we hope it’s that way, and that they don’t choose their platform on usual vendors techniques like getting paid disguised dinners, travels and hotels…”
Consider this in the context of an article clearly stating that linux has 97% share of the top 500. It seems incredibly ironic for you to say this of one of the companies in the 3%. Normally when people complain about bribery, they’re referring to the incumbents. This seemed so backwards that I had to assume (wrongly) that you were speaking of linux. Thank you for explaining it to me without attacking my education, oh wait.
Naturally
Edited 2014-06-26 00:49 UTC
Asterisk is quite a robust platform that uses Linux as a base for telecom, and there are several other Linux based solutions as well. So yeah, even Telecom is being breached by Linux, and has been for years.
Yes, Linux does have a good showing in most of those areas. Linux has succeeded where using Linux makes sense. But to suggest it’s nearly dominating them all is 1 part propaganda, 2 parts wishful thinking.
And, Linux a “premium gaming platform”? Seriously, did you manage to type that with a straight face? Hats off if so cuz that claim is hilarious.
80% of the smartphone market, near 100% smartwatch, 50% tablet, 70% server, 97% supercomputer, etc. etc. – Linux certainly seems to have a solid and growing market share or installed base pretty much everywhere I look except desktops/laptops.
Propaganda would imply I have some stake in the outcome. That’s pretty funny. What do you imagine that would be?
Of course I did.
If you are unaware of the Steam Machine, which is certainly not a low end gaming platform, you can read about it at http://en.wikipedia.org/wiki/Steam_Machine_(hardware_platform).
If you’re confusing “premium” with “successful”, try e.g., http://www.yourdictionary.com/premium. Premium actually means “at a value or price higher than normal”. The Steam Machines announced thus far are priced from $499 to over $2500; I would consider “normal” to be a PS4 or XBox One, which start around $399.
So yes, the Steam Machine is a premium platform. Perhaps you can clarify what you find so hilarious?
Linux does have a solid and growing market share in those areas. But, that’s not what you originally said. And your stake may be as simple as wanting to see Linux succeed, or wanting to see Microsoft fail.
I’m not confused about anything – certainly not about claiming Linux to be a “premium gaming platform” as silly. All your going on about Steam Machines refers to hardware only. But Linux isn’t hardware, and it’s not exactly a secret that Linux gaming is less than great to put it mildly. Seems that you fell victim to Valve’s SteamOS propaganda referring to Linux as gamings next biggest thing. That’s a future some might hope for, but it’s not reality today and isn’t forseeable that it will ever be.
Further, before you get yourself too worked up over “Steam Machines”, you should know that Alienware, one of its biggest supporters, just announced they dumped SteamOS/Linux in favor of Windows.
What I actually said was “Steam famously targeted Linux as a premium gaming platform a couple of years back”. They did, and I showed that. No idea who made the statement against which you’re trying to argue.
Actually, they didn’t.
‘According to [Alienware general manager Frank] Azor, “Valve has a lot more control developing SteamOS, ensuring it’s singularly focused with one use model. That’s why it’s a very important initiative for us, and one we’re still fully supporting as soon as it’s ready.“‘ (emphasis added)
See http://www.pcworld.com/article/2360739/meet-alienwares-alpha-consol….
Two swings and a miss. Care to try again? 😉
Edited 2014-06-26 22:32 UTC
When you use someone elses words/actions as an example in support of a point you’re trying to make, you are effectively agreeing with those words/actions. In this case the claim, which you use to support your idea of Linux domination, amounts to nothing more than propaganda ultimately designed to sell products & services.
Linux is about as premium of a gaming platform as it is as a desktop platform. In other words, people make the claim but it’s pretty far from what you see in reality.
The article I read didn’t include the quote that they still intend to support SteamOS so I’ll give you that one. My other points however are solid, which is why no credible evidence has been presented to the contrary.
I’d also like to caution you not to rely so much on quotes, announcements, press releases, and likewise. The rubber only meets the road when what you see in reality matches what you’ve heard. Simply making claims & comments doesn’t make the contents truth or fact. In other words, take what you hear with a grain of salt until there’s actual real world evidence to back it up.
Sorry, I don’t follow.
Steam selected Linux as the basis of a new, multi-vendor line of premium gaming systems (you remember, $499 to $2500ish?) that launches next year – I think you’ve finally conceded that, right?
You’re now asserting that by using that action as an example that Linux “hasn’t dominated all [product categories of real-time] quite yet”, then I must therefore “agree with that action”?
Hmmm.
Well, while I can’t follow your logic, I do in fact think that Steam’s decision to base the upcoming Steam Machines on a Linux-based OS was a good move in the long run, and the support they’ve received from multiple vendors and the extensive press coverage thus far seems to support that – IMHO, of course.
As to whether Linux can gain a significant portion of the premium gaming market, I am as always content to let the market decide. It’s been very kind to my favorite kernel this past decade. 🙂
Have yourself a great weekend!
There’s nothing to concede. Valve choosing Linux as the base for SteamOS was never in question. It’s the claim that Linux is a premium gaming platform that’s silly. Remember, we’re talking about Linux here – not hardware. It doesn’t matter what hardware Linux is running on, Linux itself is absolutely not a premium gaming platform. The market, the Linux devs, and the game makers are all in agreement with that. Linux, via SteamOS, has an incredibly difficult mountain to climb before it will ever be remotely close to a premium gaming platform.
Let’s cut right to the chase… Do you agree or disagree with Valve that Linux is a “premium gaming platform”. If you agree, please provide supporting evidence that demonstrates how Linux has earned & justified such a title.
Linux is great at certain things but being a great platform for gaming is not one of them. Not by a long shot.
If SteamOS ever becomes a success, then I may agree that it was a good move in the long run. But so far the Linux gaming track record is less than stellar to be polite. All the press coverage, press releases, and PR amounts to nothing unless something tangible follows. We’ll see.
Linux isn’t new, it has a track record in this area. I expect Linux will take gaming by storm the same way it has taken the desktop by storm. In other words, based on everything I’ve experienced thus far, I’m not holding my breath.
Thanks and you as well!
Using the dictionary definition of “premium” that I quoted earlier – “at a value or price higher than normal” – and the price points that we’ve agreed they are targeting, then of course I do.
I’ve given the justification repeatedly – the systems that Valve is targeting are significantly higher priced than normal, where normal is an Xbox One or PS4. Hence, “premium”.
I’m not sure that building higher priced, higher quality systems is something you “earn” – it’s just the market you target. Which is why I used the word “target”. See?
I can’t help but feel you’re trying to define “premium” in terms of market share, as you keep coming around to the words “dominate” and “propaganda” and “earn”. But that can’t be a valid definition. Is Android the premium platform, and iOS the pedestrian? The Camry a premium platform, and BMW the pedestrian? In both cases, it’s quite the opposite – not market share, but price and perceived (or at least advertised) value.
Or maybe you’re trying to argue Linux as a gaming platform in isolation from the other components of a system? But that makes no sense, either. What games can you play on a pure kernel, without any hardware, graphics libraries, controllers, etc.? Besides, I specified Steam’s gaming platform, not generic Linux.
Maybe you can tell me what market you think Steam is targeting with Linux, if not premium gaming?
But I doubt we’ll bridge the understanding gap here. It was a pretty simple example (one of several) where Linux doesn’t yet dominate a real-time type of market, but has an identified path toward potentially gaining some market share momentum. Not sure why that set you off. *shrugs*
Thanks for the interesting-if-sometimes-frustrating discussion. 🙂
I’m not sure why you’re back to talking about hardware. Linux is not hardware. When you describe Linux as a “premium gaming platform”, you are talking about Linux, not the hardware with which Linux is running on. You are describing the kernel and its subsystems — the software itself. That is what Linux consists of. Again, Linux is not and does not describe hardware.
If I say `Windows is a premium gaming platform`, do you think I’m describing Windows itself, or the sum of all the components that the computer running Windows consists of?
So, the issue is not redefining meanings of words (which nobody is doing), or confusion about what words mean (which nobody is), etc. The root here seems to be you originally referred to X but were actually trying to describe Y.
It’s pointless to go in circles over this so I recommend we let the numbers decide what Linux is to the gaming world. We all know Windows absolutely dominates gaming, while Linux is hardly on the map (if you can even say it is at all). Linux-based SteamOS has no showing what-so-ever at this point. Further, there’s no reason to believe Linux gaming (in any form) will ever become anything significant, much less actually dominate gaming. Saying Linux doesn’t quite yet dominate gaming is like say the teenager working the McDonalds drive-thru isn’t quite yet a millionaire.
Ack. In Germany largest export industry is automotive. A car consists up to 100 embedded systems, but most are small micro controllers for hard real time applications. Most of them are to small to run linux. Germanys second larges industry is mechanical engineering, which does less small embedded systems. But the systems has, like the automotive ones, requirements on hard real time and safety, too. Open Source Automation Development Lab (OSADL) is working to establish linux in this domain. But (unpatched) linux, at least < 3.0, is not suitable for hard real time systems.
Who knows the fraction of produced or used processors are powerful enough to run linux?
You’re right in that tiny processors can’t (and don’t need to) run Linux.
But Linux can in fact be adapted to hard real-time. Both Red Hat and SUSE offer hard real-time packages – SUSE calls theirs SUSE Linux Enterprise Real-Time, or SLERT, don’t recall the name of Red Hat’s product.
We use SLERT extensively at work to test avionics software, and it’s quite good. But it’s not appropriate for a micro-controller.
If Linux becomes the only way then evolution will have run it’s course. Are we then left with a monoculture and something approaching a blind alley?
I have a lot of time for Linux in various contexts but seriously folks… we need upstarts, experimentation, debate and competition. All of it on-going.
Since Linux is open, it’s growing dominance is not necessarily leading to a monoculture. Are Ubuntu, Android, and Chrome really the same culture? Or are they three distinct products which rely on the same kernel? Sure, I can run bash on all 3, but the primary application environments target different markets IMHO, and advance technology in different directions.
It’s possible that, should Linux attain true dominance over the industry, it will lead to even more rapid advances in technology rather than the stagnation we saw with Windows XP and Internet Explorer 6.
Hope so, at least.
Truly so. Even if Linux seems to be One, there may be very different approaches to multiprocessing one can choose (and implement). The ecosystem you fear will go missing is really built-in, to a very large extent.
If you build an extremely expensive system with a one-of-a-kind architecture, you will want your own memory manager, job dispatcher etcetera. Linux lets you add this in relatively easily, and you can even build up on previous work. I guess this is not the case with Windows, OSX, Solaris or AIX.
That’s true to a certain extent, but there is still some monoculture to be had with the kernel level. But on the whole, it is certainly a good thing to see an open operating system dominating.
Dont worry, there is no risk of monoculture..
Linux is open, and if it kills Microsoft, then be happy. This will give space to a lot of smaller open and not open projects
World would be better off with many open projects like OSV and similar stuff instead of the proprietary monoculture we got in desktop world by Microsoft.
I seriously hope Microsoft will die soon.
Has Linux’s success hurt or helped Haiku? AROS? Genode? What do you think, and why?
I think it’s helped those efforts incredibly. The projects that aren’t helped are those that are so similar in application that it’s up to smaller differences like development styles and licensing to make it worth doing something not Linux.
If Linux (the kernel) hadn’t been created by Linus Torvalds, I believe a similar product would have arisen in its place.
The hurd was the obvious heir apparent in 1990, but it lacked the practical leadership to get an actual product out the door. The obvious benefactor in hindsight would rather be BSD Unix, but I don’t believe it would have had the same level of success as Linux has seen (even in Linux’ absence), partly because Linus is a uniquely talented engineer and leader (and Linux needed both), and partly because IMHO BSD’s license offers insufficient protection from exploitation by competitors to justify the kind of investment in open code from major corporations that Linux has received. If not BSD, perhaps Minix would have overcome it’s licensing problems to become commercially viable. Others such as Haiku (fond of it though I am) would have had an even steeper playing field on which to compete than those above, but the potential is certainly there.
But I think an open kernel would have eventually and inevitably achieved widespread adoption, just not quite as successfully as Linux.
I think the amazing (to me) success of Linux has certainly had a motivational effect on the developers of later open kernels and systems, and probably a demotivational effect on developers of closed kernels and systems. Open is clearly the future IMHO. Open products are too good now, and proprietary development is too expensive, to attract the needed venture capital and industry support to proprietary kernel endeavors at this point.
But I think the most singular pivot point in history was the decision by rms to almost single-handedly create the free software movement in the 1980’s out of sheer dogged determination.
IMHO, the free software movement (and it’s nephew the open source movement) wasn’t a near-inevitable creation like the Linux kernel, but a kind of genius to see the direction that proprietary software was taking the industry, and decide that something better was needed.
Yes, he’s opinionated, controversial, prideful, lacking in personal charisma and maybe even hygiene. But people who make a real impact in the world often are. I’ve never met him, and I certainly disagree strongly with him on almost everything else, but I do respect the man for creating and executing a clear vision of a type of liberty that wasn’t at all obvious except in hindsight.
I consider myself in his debt.
That’s what I think, and why. 🙂
When I saw my favorite platform of the 80’s and early 90’s, the Amiga, put on a death march because of upper management who seemed determined to snatch defeat out of the jaws of victory, and actually showed animosity towards its enthused user base, it became important to me to invest in a platform where the enthusiasts would have say. No bigwig will ever be able to kill Linux. In fact, now that Linux has prevailed so greatly, other efforts can benefit – even the vestiges of the Amiga community with projects like AROS.
Hypervisors and microkernels get to do what they do, with L4Linux as a subsystem. Open source driver support is at critical mass for a wide range of hardware, so the door is now open for any other OS to port that over. It’s up to engineers who care about doing what they know can be done, not MBAs.
It used to be up to upper management types that cared about market share in their specific business plan. Now there’s a way to support all sorts of cool and unusual projects. Without widely supported open source operating systems, where would Arduino and Raspberry Pi be?
Linux has been very good for the whole computing ecosystem.
Edited 2014-06-25 14:29 UTC
There were other reasons …for example, the tight coupling of Amiga hardware and software meant that upgrades of the architecture (maintaining compatibility) were difficult.