Well, it started when i asked myself the question: What did Windows 3.x have when it came out that made it so popular at that time? Now we are stuck with an over-layered Windows system and it has become hard to move to other OS. Hard to move to other OS due to the lack of new hardware support for other OS, since vendors want to support the largest market share. Not that i dislike Windows, but i am also curious why did it dominate the market. What did other OS did not have that Windows had?

I learned about computers when Windows 98 was around. Therefore i don't have much knowledge in computer history. I Googled around and found many interesting articles and stories. Among all of them, this one is my favorite:  http://www.openwatcom.org/index.php/Exploring_Windows_3.x

I like that article because it is not biased on praising Windows or any specific OS or vendor. Most of the facts and flavors are still in there. And what i understood from there was that Unix systems where horrible for average users because they lacked of GUI interface. OS/2 did not take over Windows because it were not backward compatible with DOS legacy software; even though OS/2 was ahead of the game in terms of stability.

I also remember once my English/Computer teacher told me that he first became hooked into computers with the release of Windows 3.x; because it was easy. Which explains why most of the Unix-like systems faded into the background for server computers. And then it took so long for Linux variations like Ubuntu to come into the average computer user. But then it makes me wonder why did MacOS never dominated the average computer user? I mean they were innovators of the GUI and mouse interface.

Maybe some of you guys can enlighten me on that one. Did they get left behind with intuitive interfaces since the early Window releases? It might be just me but i find it sometimes hard to get around MacOS in the current releases because only find one or two ways to perform a task. While on Windows i often find like four ways. For example: main menu, main toolbar, right click, shortcuts, wizards, etc. Or was it a factor of price? Because i know that Apple is a brand name is expensive compared to custom builds of the same price.

And talking about backward compartability, have MacOS dropped backward compartability in the early years? I know that MacOS had to drop it a few years ago when they switched from PowerPC to Intel. I am just wondering if they did something similar far in the past; which may have caused part of their lost in market share. Or maybe was it a PowerPC processor thing in dropping backward compartability? Because i know that Intel x86 architecture is the base for Windows' faithfulness with legacy softwares.

And with this, it raises questions about processors. I know that PowerPC had little userbase so their product have to be overpriced. And according to Apple advertisements when it was faithful to PowerPC, it was faster than intel's x86. So making the best processor with a small market share is very hard. But my question is, was PowerPC still faster than Intel's fastest processor before Apple switched to Intel? I know it is hard to benchmark those results since it is hard to match the other specs. And not to mention Window's overhead with its overlayered architecture. But maybe some of you guys have done benchmarks on Linux with computers of similar specs but different processors.

And my last question is about processor endianess. I always though that little vs big endian was a RISC vs CISC thing. Am i right or there are RISC and CISC processors using both variations in their architecture? And how come the representation in memory is even more confusing. fore example a 2-byte number:
-256: Decimal
-000000001b:: Big Endian
-100000000b: Little Endian
-0010h: x86 Windows Memory (Little Endian)
Posted on 2012-04-02 07:30:49 by banzemanga
I remember when Windows 3.x came out but I don't really remember using it all that much... I was definitively no Bill Gates and saw no potential in Windows 3.x and even 95 at the time.
Posted on 2012-04-02 08:54:25 by JimmyClif

Well, it started when i asked myself the question: What did Windows 3.x have when it came out that made it so popular at that time?


Not much really... Windows 3.x still had limited popularity. It wasn't until Win9x and NT 4.0 that Windows fully took over from DOS.
Windows 3.x was the first reasonable version of Windows though, and as a result, it was the first one that could really present the advantages of a GUI to the end-user, so for certain tasks, it became a nice alternative to the DOS software at the time (mostly office stuff, word processing and such is just nicer with WYSIWYG. But the go-to platform for WYSIWYG was the Macintosh in those days).


What did other OS did not have that Windows had?


'Cheap' hardware, I guess.
I think the popularity of DOS and Windows were piggybacked on the success of the IBM PC as a platform. The PC was one of very few platforms where you had various brands producing 'clones', competing on price. As a result, PCs prices came down reasonably quickly, where other machines, such as the Macintosh remained expensive.
Perhaps the fact that the PC was a platform from IBM also had something to do with it. IBM was a big name in the early days of computing, and business machines in general (typewriters too). There used to be a saying something like: "Nobody ever got fired for buying IBM". Apple? That was just some newfangled company started by some strange hippies in their garage!

And what i understood from there was that Unix systems where horrible for average users because they lacked of GUI interface.


UNIX did have a GUI, in fact, it was pretty early into the game. The X Window system (which is still used by most *nix-like OSes today, such as linux, *BSD, Solaris etc) started in 1984, even before Windows.
The problem with UNIX is that it was not designed for personal computers. Your average desktop computer just wasn't powerful enough for it. DOS, Windows, classic MacOS, and various other early Personal/Home Computer OSes were far more lightweight than UNIX.
Windows NT had the same problem really: The first versions, around 1992, could only run on very high-end 386 or 486 machines, with a lot of memory and large drives. NT offered more or less the same features as UNIX, but as a result, it also required similarly powerful systems. Both Windows NT and *nix-like OSes didn't really take off on regular PCs until the mid-90s, when Pentiums became commonplace, which were powerful enough to run these OSes without a problem.

OS/2 did not take over Windows because it were not backward compatible with DOS legacy software; even though OS/2 was ahead of the game in terms of stability.


OS/2 actually did have DOS and Windows compatibility... Problem is: there was very little native OS/2 software. As a result, people would mainly use OS/2 in DOS/Windows compatibility mode anyway... so why not just use Windows itself?
OS/2 and Windows NT are actually close cousins anyway. MS and IBM started working on the OS together. But at some point IBM wanted to go the OS/2 way, where MS figured it would be better to make the OS into a Windows. Obviously, Microsoft called it correctly.


But then it makes me wonder why did MacOS never dominated the average computer user? I mean they were innovators of the GUI and mouse interface.


Price mainly, I suppose... Then as a result of that, the Mac was always a niche platform, and less software/games/etc was available.
In the early days, C64 was most popular, because it was cheap and it had tons of software. Its successor, the Amiga, was perhaps one of the most technically advanced personal/home computer platforms ever, but it never became anywhere near as popular as the C64. It was already struggling against the PC's popularity (although PCs were a lot more expensive, people were familiar with PCs from work, and there was more software available).


And talking about backward compartability, have MacOS dropped backward compartability in the early years? I know that MacOS had to drop it a few years ago when they switched from PowerPC to Intel.


Yes and no.
The Macintosh started with Motorola 68000 processors. When they moved to PowerPC, they built an emulator into the OS, so you could still seamlessly run 68k software.
Likewise, for the transition to Intel, Apple built in another emulator for PPC software (called Rosetta). They also created a system called a 'universal binary', where a single binary file could contain both PPC and Intel code, so it would work on any Apple, regardless of the CPU it used.

So yes, Apple did put in the effort to make older programs work.
On the other hand, Apple was never all that good with maintaining compatibility with OS updates. So newer OSes may break some older apps (not quite like how MS does it, where they build specific workarounds for buggy apps into the OS, see http://blogs.msdn.com/b/oldnewthing/)

But my question is, was PowerPC still faster than Intel's fastest processor before Apple switched to Intel?


No, the problem was... Originally PowerPC was a project of 3 companies: Apple, IBM and Motorola.
IBM used the POWER architecture for their servers, and wanted a scaled-down version of it for smaller workstations and desktop/laptop systems, which would become the PowerPC architecture.
IBM would make the chips for their own uses, and Motorola would be building their own PowerPC variations for Apple.

But at a certain point, Motorola gave up on the PowerPC. It cost too much to keep developing new cutting-edge PowerPC CPUs, and Apple was not selling enough of them.
The G4 was the last Motorola PowerPC.
Then Apple had to switch to IBM, the only remaining PowerPC manufacturer. But IBM lost interest not much later... So Apple no longer had an upgrade path.
In theory, an up-to-date PowerPC may still have been competitive with x86, but since no such chip had been developed in years, x86 was miles ahead by now.

And my last question is about processor endianess. I always though that little vs big endian was a RISC vs CISC thing. Am i right or there are RISC and CISC processors using both variations in their architecture?


It's not a RISC vs CISC thing at all.
There are plenty of CISC CPUs which are big endian as well (such as the Motorola 68k). Likewise, there are RISC CPUs that are little endian.
The PowerPC supported both, it could be switched to either mode. However, Apple always used them in big endian mode, probably because of their big endian heritage of the 68k.


And how come the representation in memory is even more confusing. fore example a 2-byte number:
-256: Decimal
-000000001b:: Big Endian
-100000000b: Little Endian
-0010h: x86 Windows Memory (Little Endian)


Where is the confusion?
If you have 256, in a hexadecimal word that is 0100h (Endianness only applies to how it is stored in memory, not how it is written in source code, or how it is stored in a register).
Which is a 01h high byte and a 00h low byte.
Big Endian writes the bytes from high to low, in increasing memory positions (much like how humans write decimal numbers):
01h 00h
00000001b 00000000b
Little Endian writes them from low to high:
00h 01h
00000000b 00000001b
Posted on 2012-04-02 09:02:53 by Scali
Thank you scali for so much of the debunking. Thanks to you, when i tell the history of computers of my grandkids; i won't be telling fairy tales. lol.

Here is a summary from what i understood from you:
-The reason why Unix-like system did not take over the desktop market was because it was too heavy of a load for desktop hardware.
-MacOS did not take over because there was no hardware competitors on the PowerPC RISC counterparts so they were expensive machines. Which lead to little userbase and developers.
-Backward compartbility was not the main reason why Windows soared in the game. Each OS had their own way to make legacy software work.
-Intel is now ahead of PowerPC thanks to the market-share and competition. Sadly it is currently not getting much of it from AMD.

?-When you said IBM did it the OS/2 way, i think you meant little user friendly graphical utility?
?-As for the endianess problem, i think that my example was not  clear enough. Here is another 2-byte example.
12 34h: little endian
0001 0010 0011 0100b: little endian
0010 1100 0100 1000b: big endian
2C 48h: big endian
12 34h: big endian  memory
34 12h: x86 Windows memory (little endian)

Rantings: Interesting seeing that Unix-like system lost market share due to the heavy load. The latest Windows system still runs fine (including Vista, except when it first came out) in the hardware of average users. Linux is a lot more lightweight than Windows but because there isn't a single owner to partner with hardware companies, they don't have driver support for the latest hardware. Not to mention that each Linux distribution has its own flavor so making drivers for drivers for each distribution is annoying.
Posted on 2012-04-02 15:02:32 by banzemanga

-The reason why Unix-like system did not take over the desktop market was because it was too heavy of a load for desktop hardware.


Well, yes and no... UNIX was a system for minicomputers, not microcomputers (generally you'd have one large server, and a bunch of thin clients, not desktop machines). There has been the odd UNIX-derivative aimed at regular PCs (such as Microsoft XENIX), but they were very much niche-OSes. Not many people used them, and there wasn't a lot of software for them.
In general, UNIX never even *tried* to take over the desktop market.


-MacOS did not take over because there was no hardware competitors on the PowerPC RISC counterparts so they were expensive machines. Which lead to little userbase and developers.


Not just the CPU, but the hardware in general (custom chipsets etc). There has been the odd Mac clone over the years, but Apple sued every clone maker successfully.
IBM also tried to sue the first clone-maker by the way (Compaq), but they lost the case, since IBM just used off-the-shelf hardware which any company could buy... The only thing IBM had any IP of, was the BIOS, but Compaq had cloned it via clean-room reversing and reimplementing, which did not violate IBM's copyright.
As a result, the court ruled that Compaq's clones were legal. The rest is history.
Important fact is also that IBM did not own the rights to the OS. Microsoft was smart enough to give IBM a non-exclusive license. This meant that Compaq could also license MS-DOS, which ensured 100% software compatibility.


-Backward compartbility was not the main reason why Windows soared in the game. Each OS had their own way to make legacy software work.


Not in the early days, no. These days there's so much Windows software around, that virtually any OS that can't run Windows software is doomed to fail.


-Intel is now ahead of PowerPC thanks to the market-share and competition. Sadly it is currently not getting much of it from AMD.


Yup. Intel beat every competitor with brute force, including AMD. Nobody can keep up with the rate at which Intel develops new architectures and manufacturing technologies.


?-When you said IBM did it the OS/2 way, i think you meant little user friendly graphical utility?


Not really... OS/2 actually did have quite a nice GUI. The thing is just that they chose to make it a new OS, where Microsoft branded it as Windows, and gave Windows NT the same look as Windows 3.x for DOS.
Under the bonnet, Windows NT still 'emulated' Win16/DOS in a VM, much like OS/2.


?-As for the endianess problem, i think that my example was not  clear enough. Here is another 2-byte example.
12 34h: little endian
0001 0010 0011 0100b: little endian
0010 1100 0100 1000b: big endian
2C 48h: big endian
12 34h: big endian  memory
34 12h: x86 Windows memory (little endian)


Doesn't look correct to me. You are reversing the bit order, it seems. Endianness only affects the *byte* order.


Rantings: Interesting seeing that Unix-like system lost market share due to the heavy load.


They didn't lose market share. They never had any marketshare to speak of in the microcomputer/desktop/PC market. UNIX has never been as popular as it is today with linux, BSD and OS X, because it can finally ride along on the success of the x86 platform.


The latest Windows system still runs fine (including Vista, except when it first came out) in the hardware of average users. Linux is a lot more lightweight than Windows but because there isn't a single owner to partner with hardware companies, they don't have driver support for the latest hardware. Not to mention that each Linux distribution has its own flavor so making drivers for drivers for each distribution is annoying.


I don't consider linux any more lightweight than Windows. On the contrary. The way GUIs are implemented on linux is very bloated and inefficient.
Posted on 2012-04-02 15:53:44 by Scali

Not in the early days, no. These days there's so much Windows software around, that virtually any OS that can't run Windows software is doomed to fail.

But if that is the case then, if they had those Windows tools on another OS and better+cheaper, wouldn't that have made Windows obsolete? But of course, cheapness was never the case for Mac computer.


Not really... OS/2 actually did have quite a nice GUI. The thing is just that they chose to make it a new OS, where Microsoft branded it as Windows, and gave Windows NT the same look as Windows 3.x for DOS.
Under the bonnet, Windows NT still 'emulated' Win16/DOS in a VM, much like OS/2.

Still a little confused. Does it mean that OS/2 never competed on the desktop market? But only on the server market since you are comparing it to Windows NT.


Doesn't look correct to me. You are reversing the bit order, it seems. Endianness only affects the *byte* order.

Now i see the root of the problem. I was taught endianess on the *bit* perspective. So, i always thought that endianess only applied to bit representation of numbers.


I don't consider linux any more lightweight than Windows. On the contrary. The way GUIs are implemented on linux is very bloated and inefficient.

But isn't that because Windows has the advantage of hardware acceleration? I remember that in the XP days, Linux with transparency would run faster XP with classic theme. Not sure in the new Linux distros since they don't use alpha blending as much as Windows 7. But even so, Ubuntu seems to run more smooth than Windows 7 without alpha blending in my virtual box.

Edit: I liked how you brought the lawsuit cases. I wonder how things would have turned out if MacOS did not sue against clone makers. Maybe we could be having a Mac desktop everywhere in that case. Or a tight competition against Microsoft and Windows would be a lot of cheaper.
Posted on 2012-04-02 16:37:06 by banzemanga

But if that is the case then, if they had those Windows tools on another OS and better+cheaper, wouldn't that have made Windows obsolete? But of course, cheapness was never the case for MacOS.


Well yes... The first OS to reach 'critical mass' would win. In the early days, things weren't as black-and-white as today.
In the early days of DOS, there wasn't a software advantage. All systems had their own part of the market, and had their own share of software.
Likewise, Windows didn't start with a whole lot of software either. There was the fact that it ran on DOS, and many DOS applications could be started from within Windows... but if you weren't using a Windows application regularly, you just wouldn't start Windows in the first place, and just stick to DOS.

The Mac was the first with a decent GUI, and as a result, the first WYSIWYG word processors/desktop publishing suites were on Mac. PhotoShop was also a Mac-only affair at first. As a result, the Mac cornered that part of the market, and is still often chosen over Windows PCs in these industries (even though nearly all popular Mac software has had Windows versions for years now).


Still a little confused. Does it mean that OS/2 never competed on the desktop market? But only on the server market since you are comparing it to Windows NT.


Well, yes and no...
OS/2 and Windows NT were/are multifunctional OSes. You can use them for servers and workstations, and also desktops.
OS/2 was the default OS for IBM's PS/2 line, and as such it was very much aimed at (high-end) desktop usage. But it just wasn't very popular, so not a lot of OS/2 software was ever developed. Most people either just installed DOS/Windows on their PS/2 systems, or at least ran mostly DOS/Windows apps from OS/2.
Outside of the IBM PS/2 systems, not many people used OS/2 in the first place (most PCs came with DOS preinstalled, later also Windows).


But isn't that because Windows has the advantage of hardware acceleration?


Not really. The problem is mostly that linux has layer upon layer of abstractions.
Firstly, because of UNIX' minicomputer heritage, X is designed to run over a network.
So there's a server-part and a client-part, and they communicate with eachother via a socket.
X is just the raw protocol. On top of that, you need to run a window manager.


But even so, Ubuntu seems to run more smooth than Windows 7 without alpha blending in my virtual box.


VMs are generally not a good indication of how a UI runs on actual hardware, because you lose some, if not all acceleration features of your hardware.


Edit: I liked how you brought the lawsuit cases. I wonder how things would have turned out if MacOS did not sue against clone makers. Maybe we would be having a Mac desktop everywhere in that case.


I think so. It's pretty clear that x86 and DOS/Windows did not win because they were the first to the party, or the best possible option.
I think it's a combination of the right price, and the right software... and perhaps also partly the image.
I was quite the Commodore fan in the early years. Loved my C64 (excellent GUI on there too, in the form of Geos... Incredible what they could do with just 64K of memory and a 1 MHz CPU) and my Amiga.
Especially the Amiga was very impressive at the time, in every way. It had great graphics and sound, and the OS offered a good GUI (much better than the Windows of that era), and with pre-emptive multitasking (which Windows didn't get until 95, and Mac OS not until OS X, more than a decade later).
It was also much cheaper than a PC to boot. But somehow people didn't take it seriously. They thought it was just a game machine, and wouldn't be good for business (even though things like WordPerfect did exist for Amiga).
I think if technical qualities had anything to do with it, the Amiga would have won.
Posted on 2012-04-02 17:00:46 by Scali

Especially the Amiga was very impressive at the time, in every way. It had great graphics and sound, and the OS offered a good GUI (much better than the Windows of that era), and with pre-emptive multitasking (which Windows didn't get until 95, and Mac OS not until OS X, more than a decade later).
It was also much cheaper than a PC to boot. But somehow people didn't take it seriously. They thought it was just a game machine, and wouldn't be good for business (even though things like WordPerfect did exist for Amiga).
I think if technical qualities had anything to do with it, the Amiga would have won.

I always see so much praise for the Amiga. I have read a lot that publishing and advertising was their weak point which made people not take it seriously. I thought that Amiga was history but i just found out it is not. What about the current Amiga builds? Are they good. Everyone keeps talking about it as if it is dead.


I was quite the Commodore fan in the early years. Loved my C64 (excellent GUI on there too, in the form of Geos... Incredible what they could do with just 64K of memory and a 1 MHz CPU) and my Amiga.

That reminds me of another question. I know that the x86 architecture had to come up with many addressing modes in order to accommodate backward compartbility on legacy programs. Which is why there is real mode, protected mode, etc. Does the PowerPC has the same problem with the different addressing mode? Or it is just a x86 issue?
Posted on 2012-04-02 17:31:40 by banzemanga

I always see so much praise for the Amiga. I have read a lot that publishing and advertising was their weak point which made people not take it seriously. I thought that Amiga was history but i just found out it is not. What about the current Amiga builds? Are they good. Everyone keeps talking about it as if it is dead.


Amiga has been dead since ~1994 when Commodore went bankrupt.
Since that time, various companies have owned the rights to the Commodore and/or Amiga brands, and have used them for various products.
The current 'Commodore USA' basically builds x86 machines with a customized linux distro on it ('OS Vision'), and puts the Amiga logo on the case.
It has nothing to do with the original Amigas at all.
Then there's the 'AmigaOne', which is slightly more related, but still not very.
It's a PowerPC-based system that can run 'MorphOS'.
There is a very vague relation between PPC and Amiga, since Amiga used Motorola 68k processors, and the PPC was also made by Motorola. At some point, there were PPC-based accelerator boards on the market for real Amiga systems.
There has never been any real Amiga that was sold with a PPC processor though.
Likewise, MorphOS is related to AmigaOS. It is more or less a PPC-port of the real AmigaOS. However, since there never was a real PPC Amiga, no real Amiga ever ran MorphOS either.
Also, the Amiga had a lot of custom hardware for its graphics and sound. Since the modern 'Amiga' systems don't have any of this custom hardware, they aren't 100% compatible with a real Amiga either, and can only run real Amiga software through software emulation. Which makes them about as real as a Windows PC with an Amiga emulator installed...
So yea, Amiga is very, very dead.


That reminds me of another question. I know that the x86 architecture had to come up with many addressing modes in order to accommodate backward compartbility on legacy programs. Which is why there is real mode, protected mode, etc. Does the PowerPC has the same problem with the different addressing mode? Or it is just a x86 issue?


It's x86-specific.
With the 68k for example, the CPU was designed as a 32-bit system from the start, even though the original 68000 was only 16-bit internally. So once newer, true 32-bit 68k CPUs came out, nothing needed to change.
Likewise, the PowerPC was defined from day 1 to be both a 32-bit and 64-bit platform (it was based on the POWER architecture, which was already 64-bit at that time).
The 64-bit mode was only implemented by a few IBM processors though, so most of the time, PPC appeared to be 32-bit (the only 64-bit Apple systems are based on the IBM G5). But since PPC was designed to work that way, they didn't have to massively overhaul the CPU to go from 32-bit to 64-bit.
Posted on 2012-04-02 17:44:34 by Scali
Thank you for all the clarifications scali. I can't help to look back and see how dangerous it is having all these misunderstanding is place. I am glad i asked those questions.
Posted on 2012-04-02 18:34:46 by banzemanga
It's interesting... and sad in a way... that some people have no concept of the computing world prior to the dominance of x86/Intel and Microsoft.
Heck, it seems that more and more people have never even seen a version of Windows prior to XP.
People seem scared that when AMD goes bankrupt, Intel will increase prices.
Well, boohoo... AMD has only been in the x86 game since 1991. There have been some other niche x86 clones, from companies like Cyrix and NEC, but nothing significant. Guess what? Intel still developed new CPUs, and people could still afford them, even without a direct competitor!
Even in its best days, AMD only had somewhere between 30-35% marketshare. I believe they are currently under 20%, and still dropping.

As for software... well, I mostly played games on my C64 and Amiga back in the day... but in the 80s and early 90s, there were more games released on C64 and Amiga than on PC, and they were generally better than the PC games as well.
PCs just weren't very good at that type of game, generally involving scrolling and sprites and such. C64/Amiga had special hardware for that. But then came games like Wolfenstein 3D and Doom... Amiga could not keep up with that, because Commodore had not done enough to push the platform technology. The classic Amigas were great machines in the mid-80s (A1000, A500/2000), but the updated A1200/4000 models were too little, too late when they were introduced around 1993. The A1200 was not powerful enough to do 3d games like Wolfenstein, and the A4000 was too expensive.
From that point on, PC took over as a gaming platform, and to this day, most games are 3D first-person shooters, based on Wolfenstein/Doom/Quake.
Posted on 2012-04-03 09:34:39 by Scali