• Please review our updated Terms and Rules here

Weird Intel Windows 3.1

Intel certainly made and sold machines. After all they made motherboards, so it wasn't all that hard to pull together a PC after that. I don't believe they ever sold them in retail outlets. But they sold them to businesses. The ones I have are pretty generic, so I'm sure they sourced things like the case from a third party. The power supplies are delta. But they have intel stickers and part numbers all over like you would see on major vendors.
 
I spent a Saturday night enjoying the nostalgic install of this intel 3.1 windows on my IBM AT 5170 with an Intel Inboard 386/AT + piggyback (3MB) that I fixed and an above board 8 plus/IO with 8MB
The result is beautiful fully working windows 3.1 on this AT, not much bugs beside a small issue was resolve by editing config.sys.
I will share some photos, enjoy!
 

Attachments

  • Snapchat-171437437.jpg
    Snapchat-171437437.jpg
    1.4 MB · Views: 27
  • Snapchat-1323005750.jpg
    Snapchat-1323005750.jpg
    2.3 MB · Views: 26
  • Snapchat-1997171126.jpg
    Snapchat-1997171126.jpg
    945.6 KB · Views: 26
  • Snapchat-4571530.jpg
    Snapchat-4571530.jpg
    1.3 MB · Views: 24
  • Snapchat-1805377045.jpg
    Snapchat-1805377045.jpg
    1 MB · Views: 22
  • Snapchat-1266060811.jpg
    Snapchat-1266060811.jpg
    961.1 KB · Views: 20
  • Snapchat-398618439.jpg
    Snapchat-398618439.jpg
    928.9 KB · Views: 22
  • Snapchat-2065636757.jpg
    Snapchat-2065636757.jpg
    905.7 KB · Views: 26
  • Snapchat-704259431.jpg
    Snapchat-704259431.jpg
    2 MB · Views: 26
  • Snapchat-1843780782.jpg
    Snapchat-1843780782.jpg
    1.1 MB · Views: 27
I spent a Saturday night enjoying the nostalgic install of this intel 3.1 windows on my IBM AT 5170 with an Intel Inboard 386/AT + piggyback (3MB) that I fixed and an above board 8 plus/IO with 8MB
The result is beautiful fully working windows 3.1 on this AT, not much bugs beside a small issue was resolve by editing config.sys.
I will share some photos, enjoy!

Looks like "generic" Win3.1. Since it installed Minesweeper instead of Reversi, the graphics they used on the box may just have been from a beta version or possibly a hodgepodge of both 3.0 and 3.1. The fonts they used and a couple of icons also look a little odd on the box graphics.
 
"Not for retail sale except with an Intel i386/i486 Microprosessor-based system".

Does this imply some special retail Windows pricing (and packaging) for customers choosing an Intel CPU for their new system back in 1993?
 
Still seems like oem copies for machines intel sold, if you wanted to add in windows I guess, based on what the additional pictures show. These boxes prominently display intel, and they even have the intel eula. I'm not sure why one would want to bundle a specially manufactured branded full box for intel just with a cpu at retail in those days. That would have taken effort, and if you wanted a box copy, there likely would have been boxes of windows available at the local store. If intel wanted to bundle copies with cpus for a marketing deal with Microsoft, then they'd just get a standard distribution of windows and sell that and save some pennies of their own. The stores / generic computer vendor can just get their copies the same from Microsoft if they also wanted to do the bundling gig. They did not need to approach intel for that.
 
Yes, it seems too much for just a CPU-only deal. From what I remember from the early 90s and the local stores, clone makers just sold the bare minimum: Media + Certificate + EULA wrapped on a piece of cardboard. No box.

I'll scan the EULA, perhaps the legal text will provide some hint. Or not...
 
Apparently intel! ;) Although I think it's just a misprint.

I was expecting more of 1.2M. Pretty sure it was in '92 my grandmother bought a 486 system with 5.25" HD capable drive as an upgrade path. They were still being sold new... and I understand, the media was quite cheap then.
 
"Not for retail sale except with an Intel i386/i486 Microprosessor-based system".

Which means there's likely an "Intel-branded" version of MS-DOS out there somewhere too. Would be interesting to see the box artwork and disks on that as well...
 
Windows 3.1 on 1.2mb disks has 7 disks.

It looks like rather cheap, sloppy OEM packaging.

I'd still suggest making disk images, and comparing the contents to a another vanilla or OEM Windows 3.1. Sometimes OEMs simply dropped in an extra driver file or two to support extra hardware. It is probably identical, but still good to check.
 
I wonder if those disks are really QD. 7 disks is how many were used with windows 3.1 and HD 1.2MB floppies.
I feel that's unlikely, but it certainly seems possible: 5.25" HD floppies were always 80-track 96 tpi as far as I'm aware, though 80-track double-density diskettes (720K) are exactly double the capacity of 40-track double-density diskettes (360K), which is probably what led to people calling them "quad density," whereas HD diskettes are typically 1.2 MB, which is over three times the capacity of a 40-track double density drive, making the obvious misname "hex density." (But hey, that's also "HD"! :))

Perhaps the original owner can tell us how much space is reported for those diskettes, if they're in a DOS-ish format?
 
That's a pretty interesting copy because it uses a fairly rare (in the PC world) disk format: double-density 96-tpi (i.e., 80-track) 5.25" floppies.

I know of only one PC clone (from South America, I think) that used this format. (The DEC Rainbow used 80-track diskettes, but they were only single-sided.) But surely there must have been something else out there that was more common that did.

(These are sometimes mistakenly referred to as "quad density," but they are just the same recording format as standard PC double-density (MFM) diskettes, but with twice as many tracks per side.)
There were versions of the Olivetti M21 / M24 which came with a Toshiba 96 tpi 5.25 inch double density floppy drive, sold as 640 kB drives, but instead of 8 they also can be formated with 9 sectors reaching the 720 kb format like 3,5 inch floppies.

But that is just a 8086 machine.

Can you please make images of those diskettes and provide upload them somewhere?
 
I feel that's unlikely, but it certainly seems possible: 5.25" HD floppies were always 80-track 96 tpi as far as I'm aware, though 80-track double-density diskettes (720K) are exactly double the capacity of 40-track double-density diskettes (360K), which is probably what led to people calling them "quad density," whereas HD diskettes are typically 1.2 MB, which is over three times the capacity of a 40-track double density drive, making the obvious misname "hex density." (But hey, that's also "HD"! :))

I'm also of the opinion that labeling is just "wrong" and they're normal AT-format disks. I put "wrong" in quotes there because I feel like if you're splitting hairs you can say that all the floppy formats we call "High Density" are actually just "Double Density" that happens to run at a higher data rate.

Long-winded explanation: Back in at the dawn of time "Single Density" floppy disks were a thing, and they used simple Frequency Modulated Differential Manchester Encoding, or just "FM Encoding" for short. This was reliable and easy to implement but inefficient, requiring a clock bit to be written to the media for every data bit. By the late 1970's IC density had improved enough that floppy disk controllers implementing Modified Frequency Modulation, which packs about twice as much data into the same space on the disk media without actually changing the size of the individual magnetic transitions, became practical, so reasonably quickly the entire industry moved over to these new "Double Density" controllers.

Our so-called "High Density" formats are still MFM, but they've doubled the clock rate of the controller (and thus halved the size of the magnetic transitions), but you can make a reasonable case that they're still "Double Density" formats. (Verses using some yet newer/smarter encoding system.) Both 5.25" and 3.5" "High Density" floppy controllers run at twice the clock speed of their "Double Density" counterparts, the reason the "HD" 3.5" disks hold twice as much as the "DD" version instead of the 66% more that 5.25" "HD" drives hold compared to "Quad Density" 5.25" disks is HD 5.25" drives spin at 360 RPM instead of the 300 RPM that's standard for usual 3.5/5.25" drives. (Why? Because they were designed to emulate 8" floppy drives in both RPM and data rate. It was kind of weird IBM picked it for the AT because it requires that the controller support both 300kbps and 250kbps data rates for "low density"; one for handling low density disks in 5.25" drives and the other for handling them in "real" low density disk drives or, later, 3.5" drives.) If you think about it the names we use for disk storage capacites are all nightmarishly ad-hoc and imprecise. If I took an FM controller and ran it at twice the clock speed that's usual for 5.25" disks (IE, at the speed that they run when you're driving 8" disks) it'd put the same amount of data on a disk that could handle that magnetic flux density (IE, HD media) as an MFM controller does at the lower bit speed, so why shouldn't I call that a "Double Density" Disk?

(The same arguments apply to the "quad density" name; 80 track 5.25" drives existed within a couple years of the first 40 track Shugart SA400s rolling out the door, and people did use them with FM controllers; if you're using it that way is it *also* a "Double Density" drive? I mean, it holds twice as much as the 40 track one. Also, if it's a double-sided 80 track drive it *does* hold 4 times as much as the original 40 track single-sided Shugarts, so... teamed with an MFM controller should we actually call the resulting system an "OctoDensity Drive"?) ;)

Anyway. With a special driver it's possible to read/write/format "Quad Density" floppy disks in an AT, but don't think it works without said driver? I'm pretty sure the BIOS driver in an AT figures out the data rate of a disk by spinning it and trying both data rates, and if it's readable at the lower speed it just hard assumes it also has to double-step to emulate a 40 track drive... which means reading a quad density disk would just result in sadness. But I definitely don't have a working AT at hand to try that on. If I'm *right* about that though it would seem to rule out these being quad density disks unless they were intended for a very specific machine with a BIOS that knew what to do with them.

Addendium: FWIW, the Tandy 2000 used quad density drives in a 720K format, so if you want Windows on quad density disks it exists in Windows 1.x form. But I don't suppose it really counts as a PC compatible; maybe a shade more compatible than the DEC Rainbow, but that's definitely not saying much.
 
Last edited:
I put "wrong" in quotes there because I feel like if you're splitting hairs you can say that all the floppy formats we call "High Density" are actually just "Double Density" that happens to run at a higher data rate.
I suppose that's an approach one could take, if one were focusing on the density of the encoding, but I think that's a terrible idea since you then need to entirely rework terminology for one of the most important parameters (and the one that causes the most incompatibility): the recording media.

This is why I refer to media as simply "standard density" and "high density." Any media labeled "single density," double density," or "quad density" is all standard density which, while not entirely straightforward, at least leaves the end user with a clear understanding of which media she should be using for what. (Well, there was also hard-sectored media, which fortunately fell out of fashion fairly quickly.)

Once you have media compatibility sorted out, the only other physical format compatibility issues of concern are from the drives: number of sides (one or two) and number of tracks/track density. This can be summarised as 48 or 96 tpi for 8" and 5.25" systems, and 40-track (rare) and 80-track for 5.25" and 3.5" systems. (There are some slight variants here, such as using only 35 tracks on a 5.25" diskette.)

And that's about it: the rest is on the controller and software, which of course is the same across e.g. all PC/ATs, all FM-7/FM77s, or whatever system you happen be using. There's not really a lot of point in most discussions worrying about whether the encoding used by that is FM, MFM, GCR, or whatever.

Our so-called "High Density" formats are still MFM, but they've doubled the clock rate of the controller (and thus halved the size of the magnetic transitions), but you can make a reasonable case that they're still "Double Density" formats. (Verses using some yet newer/smarter encoding system.)
So this is a case where for me, clarity reigns over accuracy: almost nobody cares that it's MFM instead of a newer/smarter encoding system. Everybody cares that you need to use HD instead of standard density media.

If you think about it the names we use for disk storage capacites are all nightmarishly ad-hoc and imprecise.
This is probably the core of the failure: not the imprecision but the fact that we're trying to use names involving "density" for disk storage capacities, which of course are a function of media, head step pitch, flux-level encoding, and sectoring. That's four different densities right there; no wonder people are confused! (And you give some nice examples of exactly this.)

So here's my suggestion about how we can all communicate more clearly about this:
  1. Use "density" to refer only to media, which are "standard density" and "high density" (HD). (Or one of the rare super-floppy densities.) In particular, never use the term "quad density."
  2. Use "single-sided" or "double-sided" as we already do.
  3. Use the number of tracks per side for track pitch. "40-track" and "80-track" cover all the most common cases (5.25" and 3.5" systems), and those using less common systems will know that, e.g., "35-track" is the same pitch as 40-track on 5.25" systems, and "77-track" is the common 48 tpi for 8" systems.
That covers all areas of physical compatibility for drives and media, including situations where you can read but not reliably write certain formats (e.g., 40-track diskettes in an 80-track drive). All the rest is up to the controller and software driving it, which can be given as a specific capacity and system/software name: e.g. "320K FM-7" or "360K IBM PC DOS."
 
I suppose that's an approach one could take, if one were focusing on the density of the encoding, but I think that's a terrible idea since you then need to entirely rework terminology for one of the most important parameters (and the one that causes the most incompatibility): the recording media.

In case it wasn’t obvious my comments about “why can’t we call *this* (several goofy examples) “double density” were mostly tongue in cheek and I’m not advocating for not calling the 1.2MB “high density”. But nonetheless I wanted to make a point here that technically a *lot* of floppy related terms are ambiguous and there are times where you can’t actually be sure if the same words always mean the same thing.

Anyway. In my opinion it remains the case that the vast majority of the evidence here suggests the use of “DS/QD” on those disk labels was just a confusing oversight/mistake on the part of whoever was running the in-house labeler at Intel. To whit:

  1. As has been observed, normal retail copies of Windows 3.1 are also seven disks. To fit the same amount of data on 720k disks would take like 12 of them.
  2. Again, I don’t have the hardware on hand to verify, but I’m pretty sure a 5170 needs a driver to read 720k disks, AND,
  3. The box these disks were in *says* it needs a “high-density” drive.
But, yes, the OP could solve this by checking the format easily enough.
 
Last edited:
  • Like
Reactions: cjs
While we're being nit-picky, Microsoft also incorrectly referred to 360K disks as "low density":

s-l1600.jpg
 
Back
Top