A friend and I used to compete on kernel size. Research the kernel features we actually used, which would be necessary during boot, etc. This was either late 2.4 kernels or early 2.6 ones. I don't remember for sure, but I think we hit about 800KB, and would've had to start adding more modules into the initrd to get smaller.
I've seen inexplicable dates show up on dead SourceForge projects before. My guess is that some of the "update" dates may represent internal changes within SourceForge, rather than intentional changes by project owners.
this is cool and all, for historical purposes. But for modern "small spec" the i386 distribution of debian, no xorg or desktop manager is probably the smallest usable distribution that has a full set of packages and will run on an ancient machine. Any old laptop that has 128MB of RAM should be capable. One of the issues you might run into is that the only way to do the install is via a burned 700MB CD-R of the ISO, and very few people actually keep CD-R media around or CD burners anymore.
At the time of course prices were at a different scale than today. My first Linux was on a 486 with I think 4 or 8mb of RAM (Slackware, install by floppy, 1998)
a little bit later than that, but I remember installing OpenBSD by 1.44MB boot floppy and then bringing up the network, retrieving the rest by the ftp tool built into the installer.
My first install was slackware 3.0 on 4mb. You had to manually partition the disk and setup swap space before you could even launch the installer.
This was a minor inconvenience compared to swapping the 40 1.44mb disks needed to install it because I couldn't get it to use my pre-atapi cdrom drive correctly. Using the cdrom drive it would work for a little bit, and then everything would start timing out. And, a network install was some science fiction as far as I and my 9600bps modem were concerned.
Then once you got it all installed you could start trying to get X11 and ppp to work...
My first install was probably Slackware 3.0 too (considering the dates, it could have been 2.3, but I don't think I've used a kernel older than the famously stable 1.2.13), but it was not on my own machine. Did that with two friends on one of them's Pentium. We had downloaded a set of 40 floppies at university and brought the whole thing to his place (because we also didn't have either access to a modem and/or a cdrom drive).
We had bought 40 floppy disks in a bundle, and it turns out one of them was crappy and ended corrupted. It also turned out it was the last of the set of disks for the X11 package. So we had a shiny Slackware install... without X.
Fortunately, iirc, Doom was there intact, and we could run it with the "console" svga backend (that was outside X, but before Linux had the framebuffer, I don't remember how that worked exactly)
Had to wait for the next day to re-download the corrupted image on a different floppy, and install the X11 package.
That must have been frustrating to say the least. I have many memories of transporting essays to school on a floppy, only to have it be corrupted by the time I got there.
Marginally off topic, but this sub-thread reminds me of the many times I have seen/heard people pining for ye olden days of computing. For whatever reason, people remember them through rose colored glasses.
They will say computers were better then. I disagree entirely. They were horrible and had much less capacity.
I see the same thing for the Internet. People will say they miss the time when it wasn't open for commercial activity. Hogwash... I know, I was there. Content was lacking, speeds were horrible, discoverability was an arcane art, and the costs were obscene.
I'm guessing, not my domain, there is some sort of psychological reason for this sort of thing. Computers were giant, more limited, and expensive. The 'net was no better.
I remember trying to research things on the web before the days of Google and Wikipedia- it was much easier to go to the library back then, in my experience. The ways that these tools have changed the ways that people work is astounding
A while ago I tried installing Slackware 3.0 on Virtual Box out of nostalgia, and it was everything I remembered and worse. I only managed to get XFree86 to start up in some 300x200 mode that would pan around the 'virtual' size. Getting networking working required a kernel recompilation.
Even after all that, there wasn't much it could do. I'm not sure why I insisted on running that instead of Windows 95.
To this day, I can remember how to build a 1.2.X kernel. You were also very lucky if you had a soundblaster card, because only a handful of sound drivers that were in the kernel.
A little later OSS came out. To get sound to work a lot of cases you had to shell out money for commercial sound drivers, or get a crack from #linuxwarez.
* You had to manually partition the disk and setup swap space before you could even launch the installer.*
As of my last install of Slackware 14.1 you still have to do this.
Slackware is a wonderful distribution and keeps most important packages pretty close to up to date but Patrick seems to very much be a "if it ain't broke don't fix it" kind of developer. My first Slackware was 2.1 or 2.2 (just remember if was 1995) and other than the fact I used an ISO image of 14.1 instead of 10's of floppies the install is very much the same after all these years.
TinyCore is just 16MB (up from the 10MB last I used it), or 11MB without a desktop environment or Xorg. But it still requires 46MB RAM to run (28 without X).
If they do still include this as a standard feature it's really a waste. Just more trash in the world. Everything can run off a USB device and really should.
Don't we have enough garbage already?
They should be optional upgrades only at this point in their life cycle.
Plus external CD+DVD burners from major brand names are now dirt cheap and run entirely off the USB port with no extra power adapter needed (!!!) so they're convenient and inexpensive to add if you really need one, with the bonus that you can move them between machines as desired.
A lot of older hardware designed for office use supports booting over ethernet using PXE[0]. You can set up a PXE server, and then configure the BIOS to "Boot over network". (Assuming it's not _so_ old it lacks an ethernet adapter)
It's a bit of a hassle, but at east it doesn't require burning physical media.
Good old days. I remember trying to shoehorn Slackware onto an IBM PS/2 m55 with 2.9mb RAM. Unluckily for me, it was an MCA system with an ESDI disk, with the MCA support very experimental, and I ended using a parport ZIP100 to help with a minimal install. One day I got back from school and some dude from one of the Usenet groups called me on my land line asking for the steps.
Argh, in college I had an old Microchannel IBM that I installed linux on. I could never get the (coax) ethernet working properly. Pretty sure I used a parallel port zip drive to install, too. Ah, the good old days of having seven computers in one tiny dorm room.
Wow, this is from a different world! When this was published, the 2.4 kernel was just a twinkle in Linus' eye--the 2.4 kernel that Damn Small Linux would later stake itself to when it boldly declared that it would never switch to 2.6 because it was just too big.
Huh. My first PC was a Cyrix "P166+" (running at 133 MHz) with 16 MB of RAM and a 1.2 GB hard drive. It was not really a high end machine at the time. The hard drive was quite big for the time, but the CPU became a liability once Quake came out. :-/
Quake ran just fine on P133. I know because I owned one at the time. Even though that game hardly did any video card magic IIRC it still was one of the major bottle necks so probably not having a good video card was your problem (I had Number 9 something or another).
I remember when winquake/opengl quake or whatever it was called came out that allowed use of the DRX/opengl drivers. It was a major improvement speed wise (as well as not having to boot into DOS and it allowed higher resolutions reliably).
However Linux Quake network wise (and startup speed) was absolutely faster than both winquake and the DOS TCP/IP stack version. A lot of people/clans that played competitively would use Linux because of this reason (as well of course the server version).
Quake was one of the things that kept me on Linux most of the time (highschool circa 1997-1999).
> Quake ran just fine on P133. I know because I owned one at the time.
In the absence of hardware acceleration, the Pentium 133 outperformed the Cyrix CPU by a wide margin when running Quake. You are correct that with hardware acceleration, the CPU was far less of an issue. But all I had was a crummy ATI Mach 64, which was not even a good 2D card. So for me, the CPU was the bottle neck.
Ironically, on integer workloads, the Cyrix was a lot faster than the counterparts made by Intel. But their FPU performance was rather underwhelming. On top of that, Quake had hand-rolled assembly code that took advantage of the Pentium's architecture.
I never knew that people used Linux primarily to play Quake. ;-)
NetBSD will install & run in less, but you need to manually customize the installation and it requires a full 486 (with FPU) at minimum. It doesn't seem like it would be that hard for someone to bring back support for no-FPU Intel again, and also to support installation with less RAM.
If you're into running GNU/Linux on older hardware, I can't recommend kmandla's blog[0] (abandoned since 2015, also skip the last post to go to the juicy bits) and Inconsolation[1] (by the same guy, abandoned since 2015 too) enough.
If it was built for 486SX, you could probably run stock NetBSD decently on this system with performance not much different from when the system was new.
Edit: Looks like you would be able to do a customized low-memory/disk install of NetBSD 7.1 on these systems if they had an FPU (either a full 80486 or an 80487 coprocessor). With 32MB RAM and 250MB disk in addition to the FPU, you wouldn't even need to do a customized install.
While we're at it, I grabbed a grid laptop (not the compass), an actual 386 based machine, is there anything recent I can run on that ? or should I just write a asm monitor and bootstrap emacs from there ?
Circa 2000 I was using a Compaq Contura 4/25cx laptop with Debian Slink, 2.0.36 kernel. It had a 486 at 25 MHz and 4 MB RAM. I also got a parallel port Ethernet adapter from the same dumpster dive. Even with Slink installing with 4 MB RAM required some special care. My notes from the time say it took one and a half minutes to boot.
There existed a driver in XFree86 for the funky Compaq AVGA graphics so I could use X, but the limited RAM made it impossibly slow to run any kind of desktop environment or even a window manager.
The setup I used on it was the zed text editor in one text console, shell for running LaTeX in another console and the gv Postscript viewer without a window manager running on a X screen. I wrote quite a lot of lab reports that way.
Documentation about that setup was one of the first web pages I made:
I installed Slackware 3.9 or 4.0 on a Contura 4/25C via floppy around the same time, but as I had 8Mb to play with I managed to get a browser running in X (though without a window manager).
I used Linux on a Toshiba T1900C (20Mhz 486SX, 4MB/170MB) [1]. I started with SLS [2], later moved to Slackware 3.0. SLS was installed through floppies which I pulled from tape on an RS/6000 7012-32T, marking the transition from 'real but expensive unix' to Linux in one fell swoop. X was a bit sluggish on the thing but for the rest it was quite usable.
Using Linux in this early age had advantages but it also came with some risks. Not risks to me as a Linux user but risks to those exposed to things I produced. I managed to create a CD which was distributed with a magazine I was editor for in that period. The CD worked fine on Linux, it worked fine on OS/2 - the other system I used when not on Linux - but it failed miserably on Windows 3.1x due to the use of lower-case filenames. This was... unfortunate... as that CD had already been pressed, what to do? I ended up solving the conundrum by distributing a tailored version of MSCDEX [3] which could handle these files, affected users could order a floppy free of charge.
Interesting story, thanks for sharing! I too started off with Linux back in the day - I'd been on minix-list for a few weeks after just getting a 386 and wanting something decent to run on it, replacing the Desqview configuration I'd been using previously. So I was quite happy to download Linus' little kernel and see what it could do .. one floppy at a time from funet.fi, lol. What a great impression it made - a real unix on a 386! Of course, the fact that it took a week to compile X/Windows wasn't that big a deal. :)
Then I moved on from building everything from scratch - a great learning experience back in the day - to Yggdrasil, which was simply a great distro for the time. Ah, good ol' days .. I really wonder what I'd do different if I had a time machine and could go back and tell myself all the things Linux would be used for in the decades ahead, it would have really blown my mind .. Still does!
`make menuconfig` still works fine, I use it to configure kernels for use on Android. Since I build those on a server somewhere on the 'net - actually it sits under the stairs but for all intents and purposes it could have been on the other side of the world - it is handy to be able to use a console-based method.
I do prefer `menuconfig` (combined with `oldconfig`) in any case, possibly out of habit but also because this type of text-based user interface often is much faster to use than its graphical equivalent.
Not really. I just don't bother building my own kernel any more - the distro I use (Ubuntu Studio) comes with one that works great out of the box. Nice to know I still could, though .. ;)
I installed Debian on my Libretto 30 ( https://en.wikipedia.org/wiki/Toshiba_Libretto - AMD 586 DX4 100 MHz, 8 MB RAM) some time in 2000. It worked pretty well - PCMCIA networking, X11 and so on. I think I used the installation technique of removing the hard disk and pre-loading Linux onto it. The machine was of course far too small to have a floppy or CD drive, and predated USB.
(I also kept the Win95 partition)
Given sufficient interest I'm sure I could produce pictures.
My 380ED was my last awesome thinkpad, I think I had potato on it. After that they went to "good" then "ok" then where they are now.
I recall trying to compile a 2.4ish kernel with support for my pcmcia network card on an old 486 sx25 (dx?) laptop. Never got it to work. The desktop was fine from redhat 6 onwards (under 5.2 my sis graphics card wouldn't work in graphics mode, so no X)
Hm, not much wrong with the T42p I'm using right now. Superb screen with the correct (4:3, 1600x1200) geometry, good keyboard, trackpoint+trackpad, still works as a daily machine even though it is 13 years old. I'd move the Thinkpad line's demise to the point where they moved away from stolid, solid working machines to follow current styles - flat keyboards, etc.
That's interesting, but do you connect those Librettos to the internet to use mIRC? Sounds a bit scary with Windows 98 in 2017. I wonder how a newer BSD or Linux would run on it with say BitchX or some similar older irc client.
Thanks for mentioning the Libretto. I had been looking for something of that form factor and age for a while and this looks like it fits the bill perfectly! Now to find one in good working condition...
I have one of those too. I picked it up at the MIT Swap Meet from an old man who sold it for me for $300 (which I thought was high at the time). That same old man was also selling another interesting piece of computing memorabilia -- a complete Enigma machine -- for about $500,000 or so. Now that's too rich for my blood!!
I managed to install DSL by connecting the HDD to another machine and using Virtualbox. In my case it was a Libretto 50CT with 24 MB of RAM: https://github.com/cfenollosa/dsl-libretto50ct
Apparently I didn't keep the Win95 partition and instead that's on a disk image "somewhere". Probably on another computer that hasn't been booted in a decade.
For my museum I have Linux (and Minix) running on some pretty old stuff. Workstations & servers (Sun/SGI/HP) still run fine with Linux/Unix on them; those laptops are quite painful if you use X.
I was running Slackware mid '90s and I remember beating my head on the table how slow it was compared to the Sun machines I used in uni with running X. But it beat Win3.1/95 already in my humble opinion so I stuck with it until now (although changed to Debian and then Ubuntu recently). Still allows me to, for most dev, to use laptops from 6-7 years old (my goto dev laptop is a Lenovo X220 from 2011 ; I bought a big stack for almost nothing last year and they are excellent).
Reminds me of using loadlin to boostrap linux from DOS because the internal CDROM didn't support boot and I didn't have the external floppy drive, and since DOS would be overwritten it would have required removing the HDD if the installation failed.
The trickiest part was getting DOS/Windows back with a custom bootable CD after I realized how slow it was with a full linux distribution.
In 2000, someone gave me an old 386 laptop, 640x480 grayscale VGA screen, probably 100MB HD, don't remember the RAM (2 or 4 MB), no ethernet. I had to find an old-for-the-time Slackware to run on it, but I got it set up, with a null-modem cable to the serial port on my desktop, and was able to use it as an X terminal. Mostly only simple Xlib apps like XTerm and XClock, but I think I may have displayed Netscape remotely on it a few times.
To think that the RAM in the smallest system I have used in my life is 256MB (somewhere in 2005 and in Windows 98 and 40GB HD) and that people used to run entire PCs in less. Feels unreal (just like people talking about 4K memory as state of the art decades ago). Nowadays people are indifferent to Memory sizes as they just keep increasing every year.
The Commodore VC-20 came with 5kb of RAM and was a "real computer".
Nowadays if you play around with Arduinos, the AVR ATmega328P on an Arduino Uno board has 2KB SRAM.
You can't really compare the ATmega328 and the VIC-20 accurately, mainly because the former is (arguably) a Harvard Architecture device.
For the most part, the cpu in the 328 doesn't have general read-write access to the 32k flash memory (used for program storage), outside of some basic functionality mainly limited to bootloader usage (at a machine code level, and IIRC only for the first block of memory), and read-only capability otherwise (so you can store data prior to loading it into flash to be read later by the code).
The 2k static RAM (SRAM) is read-write always and anywhere, anytime - and so it used mainly for variable storage. There's also a small amount of general purpose flash memory available, which is also read-write capable (but has a limited lifespan just like any flash memory - though from testing, in general you won't reach the lifespan unless you do something really stupid) that is typically used to store configuration settings and whatnot between reboots.
Whereas on the VIC-20, the CPU had complete access to all of the RAM and ROM, and full control over the address and data lines of the bus.
Each architecture has its pluses and minuses, but in general the Harvard scheme is used when you don't want the code to be able to alter itself (either intentionally or accidentally due to a buffer overrun or something), which might cause a problem that could have consequences; for an embedded microcontroller, this makes sense, as you just want the thing to run continuously and reliably on every reboot or restart, without worry that the code will corrupt itself and lead to problems.
Around 2005 or 2006 I had a Windows 98 computer with 32MB RAM in my dorm room, it was more than adequate for browsing the web and writing Ruby and C (including doing an OpenGL tutorial, NeHe's of course).
Note that according to the first google result for "2006 computer spec"[1], 1GB RAM was common at the time.
The first computer that I remember in my family was a PC Jr with either 64 or 128KB of memory. We had some variant of the Apple II at school in the computer lab (I'd guess the Apple IIe, with 64-1MB of RAM). I remember in the early 90s having a PC, with the memory going up through the single megabyte range, then into the 10s as we upgraded. Maybe 8MB in 1993 and 32MB by 1997? My parents gave me my own computer in 1999 with 96MB of RAM. I think I upgraded it to either 256 or 512MB around 2002 or 2003.
Disk storage grew along with it. Hundreds of KB with the earliest machines I used (on floppies and cartridges), hundreds of MB in the 90s (and like 1.1GB when Dad added a second hard drive to that machine!!), 6.4GB in my own first one, and so on.
It's amusing to think of my son saying things like "I've used a computer with only 4GB of RAM!" when he's older.
I recently switched to openSUSE from a Debian based background. I gotta say I've never been happier. My hardware (HP x360 2-in-1 with intel i3-6100u) is supported where I had to fiddle with some configurations in Ubuntu just to get the wifi going. It's so much faster than Windows and Ubuntu, and easy on development. Highly recommend for users looking for something slightly more advanced than Ubuntu or Debian.
While not smaller than this, the smallest (and my first distro) linux install I ever did was on a Compaq 386/SLT laptop (luggable?) that had 6 MB of RAM...
...why such a weird amount of RAM? Well, when I bought the laptop used its battery (NiCad) was dead, dead, dead - and I didn't have the money to have it properly re-built. So instead, I built my own. I scrounged a bunch of old but usable AA cell NiCad packs from cordless phones (remember those?) and soldered up my own pack, and wrapped it in a bunch of layers of electrical tape. Unfortunately, it didn't fit into the battery compartment completely...
One end was being blocked by the memory! I found that by cutting the battery pack area larger, and removing one stick of RAM (the laptop had 8 MB of RAM originally - 4 x 2 MB SIMMs) - it would fit and still boot - so that's what I did.
The laptop originally ran DOS. My first mod was to install Caldera's OpenDOS (a great version of DOS for the time, btw). Then I found and installed...Monkey Linux. This was an interesting "cut down" but fairly full-feature linux distro. It's crazy claim to fame was using the DOS filesystem for itself, instead of it having its own filesystem:
93 comments
[ 3.6 ms ] story [ 140 ms ] thread-rw------- 1 root root 7.3M May 17 16:13 vmlinuz-4.10.0-22-generic
On my machine:
The script to uncompress the kernel comes from the kernel source: https://raw.githubusercontent.com/torvalds/linux/master/scri...Arguably, not everything in that vmlinux is actually loaded:
Counting BSS:https://en.wikipedia.org/wiki/Tomsrtbt
The website is dead now. Here's the wayback machine link: https://web.archive.org/web/20160120122627/http://www.toms.n...
https://sourceforge.net/projects/mulinux/files/iso/
https://sourceforge.net/p/mulinux/wiki/browse_pages/
https://sourceforge.net/projects/mulinux/reviews
it's pretty stupid.
This was a minor inconvenience compared to swapping the 40 1.44mb disks needed to install it because I couldn't get it to use my pre-atapi cdrom drive correctly. Using the cdrom drive it would work for a little bit, and then everything would start timing out. And, a network install was some science fiction as far as I and my 9600bps modem were concerned.
Then once you got it all installed you could start trying to get X11 and ppp to work...
We had bought 40 floppy disks in a bundle, and it turns out one of them was crappy and ended corrupted. It also turned out it was the last of the set of disks for the X11 package. So we had a shiny Slackware install... without X.
Fortunately, iirc, Doom was there intact, and we could run it with the "console" svga backend (that was outside X, but before Linux had the framebuffer, I don't remember how that worked exactly)
Had to wait for the next day to re-download the corrupted image on a different floppy, and install the X11 package.
They will say computers were better then. I disagree entirely. They were horrible and had much less capacity.
I see the same thing for the Internet. People will say they miss the time when it wasn't open for commercial activity. Hogwash... I know, I was there. Content was lacking, speeds were horrible, discoverability was an arcane art, and the costs were obscene.
I'm guessing, not my domain, there is some sort of psychological reason for this sort of thing. Computers were giant, more limited, and expensive. The 'net was no better.
It happens with other things. People say cars were better. Nope. Life was better. Probably not. Etc...
Even after all that, there wasn't much it could do. I'm not sure why I insisted on running that instead of Windows 95.
My only defense is that I didn't do so out of a fake memory of a glorious past. I just like tinkering.
This was a minor inconvenience compared with recompiling the kernel overnight in order to add soundblaster support. (This was before modules.)
(Oh, and I learnt not to buy cheap no-name floppies, after the installation process failed several times, each at a different stage.)
A little later OSS came out. To get sound to work a lot of cases you had to shell out money for commercial sound drivers, or get a crack from #linuxwarez.
As of my last install of Slackware 14.1 you still have to do this.
Slackware is a wonderful distribution and keeps most important packages pretty close to up to date but Patrick seems to very much be a "if it ain't broke don't fix it" kind of developer. My first Slackware was 2.1 or 2.2 (just remember if was 1995) and other than the fact I used an ISO image of 14.1 instead of 10's of floppies the install is very much the same after all these years.
You still do with Slackware ;)
Don't we have enough garbage already?
They should be optional upgrades only at this point in their life cycle.
It's a bit of a hassle, but at east it doesn't require burning physical media.
[0]https://en.wikipedia.org/wiki/Preboot_Execution_Environment
Still, it's tough to fit in 256MB with a modern kernel and Xorg.
I remember when winquake/opengl quake or whatever it was called came out that allowed use of the DRX/opengl drivers. It was a major improvement speed wise (as well as not having to boot into DOS and it allowed higher resolutions reliably).
However Linux Quake network wise (and startup speed) was absolutely faster than both winquake and the DOS TCP/IP stack version. A lot of people/clans that played competitively would use Linux because of this reason (as well of course the server version).
Quake was one of the things that kept me on Linux most of the time (highschool circa 1997-1999).
In the absence of hardware acceleration, the Pentium 133 outperformed the Cyrix CPU by a wide margin when running Quake. You are correct that with hardware acceleration, the CPU was far less of an issue. But all I had was a crummy ATI Mach 64, which was not even a good 2D card. So for me, the CPU was the bottle neck.
Ironically, on integer workloads, the Cyrix was a lot faster than the counterparts made by Intel. But their FPU performance was rather underwhelming. On top of that, Quake had hand-rolled assembly code that took advantage of the Pentium's architecture.
I never knew that people used Linux primarily to play Quake. ;-)
[0] https://kmandla.wordpress.com/
[1] https://inconsolation.wordpress.com/
I can heartily second the recommendation, and I'd love to know what the author is up to nowadays.
I'm thinking it results in a basic OS layer that provides access to the 'standard' I/O (parallel, serial, vga, hdd) ?
Much like a box that only has Dos 6.22 on it ?
Edit: Looks like you would be able to do a customized low-memory/disk install of NetBSD 7.1 on these systems if they had an FPU (either a full 80486 or an 80487 coprocessor). With 32MB RAM and 250MB disk in addition to the FPU, you wouldn't even need to do a customized install.
There existed a driver in XFree86 for the funky Compaq AVGA graphics so I could use X, but the limited RAM made it impossibly slow to run any kind of desktop environment or even a window manager.
The setup I used on it was the zed text editor in one text console, shell for running LaTeX in another console and the gv Postscript viewer without a window manager running on a X screen. I wrote quite a lot of lab reports that way.
Documentation about that setup was one of the first web pages I made:
https://www.tablix.org/~avian/contura/
Using Linux in this early age had advantages but it also came with some risks. Not risks to me as a Linux user but risks to those exposed to things I produced. I managed to create a CD which was distributed with a magazine I was editor for in that period. The CD worked fine on Linux, it worked fine on OS/2 - the other system I used when not on Linux - but it failed miserably on Windows 3.1x due to the use of lower-case filenames. This was... unfortunate... as that CD had already been pressed, what to do? I ended up solving the conundrum by distributing a tailored version of MSCDEX [3] which could handle these files, affected users could order a floppy free of charge.
[1] http://www.computinghistory.org.uk/det/13145/Toshiba-T1900C/
[2] https://github.com/rdebath/SLS-1.02
[3] https://en.wikipedia.org/wiki/MSCDEX
Then I moved on from building everything from scratch - a great learning experience back in the day - to Yggdrasil, which was simply a great distro for the time. Ah, good ol' days .. I really wonder what I'd do different if I had a time machine and could go back and tell myself all the things Linux would be used for in the decades ahead, it would have really blown my mind .. Still does!
But .. nah.
;)
I do prefer `menuconfig` (combined with `oldconfig`) in any case, possibly out of habit but also because this type of text-based user interface often is much faster to use than its graphical equivalent.
(I also kept the Win95 partition)
Given sufficient interest I'm sure I could produce pictures.
I recall trying to compile a 2.4ish kernel with support for my pcmcia network card on an old 486 sx25 (dx?) laptop. Never got it to work. The desktop was fine from redhat 6 onwards (under 5.2 my sis graphics card wouldn't work in graphics mode, so no X)
Its going to be dire.
My Amiga's and DOS machines are online too.
Unfortunately the Transmeta CPU in that thing was really slow.
Apparently I didn't keep the Win95 partition and instead that's on a disk image "somewhere". Probably on another computer that hasn't been booted in a decade.
I was running Slackware mid '90s and I remember beating my head on the table how slow it was compared to the Sun machines I used in uni with running X. But it beat Win3.1/95 already in my humble opinion so I stuck with it until now (although changed to Debian and then Ubuntu recently). Still allows me to, for most dev, to use laptops from 6-7 years old (my goto dev laptop is a Lenovo X220 from 2011 ; I bought a big stack for almost nothing last year and they are excellent).
https://www.coreboot.org/Board:lenovo/x220#Flashing
The trickiest part was getting DOS/Windows back with a custom bootable CD after I realized how slow it was with a full linux distribution.
For the most part, the cpu in the 328 doesn't have general read-write access to the 32k flash memory (used for program storage), outside of some basic functionality mainly limited to bootloader usage (at a machine code level, and IIRC only for the first block of memory), and read-only capability otherwise (so you can store data prior to loading it into flash to be read later by the code).
The 2k static RAM (SRAM) is read-write always and anywhere, anytime - and so it used mainly for variable storage. There's also a small amount of general purpose flash memory available, which is also read-write capable (but has a limited lifespan just like any flash memory - though from testing, in general you won't reach the lifespan unless you do something really stupid) that is typically used to store configuration settings and whatnot between reboots.
Whereas on the VIC-20, the CPU had complete access to all of the RAM and ROM, and full control over the address and data lines of the bus.
Each architecture has its pluses and minuses, but in general the Harvard scheme is used when you don't want the code to be able to alter itself (either intentionally or accidentally due to a buffer overrun or something), which might cause a problem that could have consequences; for an embedded microcontroller, this makes sense, as you just want the thing to run continuously and reliably on every reboot or restart, without worry that the code will corrupt itself and lead to problems.
Note that according to the first google result for "2006 computer spec"[1], 1GB RAM was common at the time.
Oh, Firefox 1.0, we miss you so!
[1] http://vbcity.com/forums/t/133493.aspx
Disk storage grew along with it. Hundreds of KB with the earliest machines I used (on floppies and cartridges), hundreds of MB in the 90s (and like 1.1GB when Dad added a second hard drive to that machine!!), 6.4GB in my own first one, and so on.
It's amusing to think of my son saying things like "I've used a computer with only 4GB of RAM!" when he's older.
...why such a weird amount of RAM? Well, when I bought the laptop used its battery (NiCad) was dead, dead, dead - and I didn't have the money to have it properly re-built. So instead, I built my own. I scrounged a bunch of old but usable AA cell NiCad packs from cordless phones (remember those?) and soldered up my own pack, and wrapped it in a bunch of layers of electrical tape. Unfortunately, it didn't fit into the battery compartment completely...
One end was being blocked by the memory! I found that by cutting the battery pack area larger, and removing one stick of RAM (the laptop had 8 MB of RAM originally - 4 x 2 MB SIMMs) - it would fit and still boot - so that's what I did.
The laptop originally ran DOS. My first mod was to install Caldera's OpenDOS (a great version of DOS for the time, btw). Then I found and installed...Monkey Linux. This was an interesting "cut down" but fairly full-feature linux distro. It's crazy claim to fame was using the DOS filesystem for itself, instead of it having its own filesystem:
http://projectdevolve.tripod.com/