I remember all those video game boxes stamped with IBM PC (and compatible), while bringing good memories I felt this was rather incorrect.

Most of these games would never run on such computers, i.e. they display 256 colors, output audio on sound cards and most importantly, have much higher CPU requirements; technologies that surfaced during AT era.

Since the natural descendants of the PC are the XT then the AT, I believe that our PCs are in fact ATs, is that correct ?

  • 2
    What we "commonly" call PCs includes Apples, Wangs, Acorns, Linux boxes, etc, etc. But yes, the product called "IBM PC" was only made for a short time.
    – Chenmunka
    Mar 21, 2018 at 9:10
  • 6
    The AT is the IBM PC AT, just as the XT is the IBM PC XT, so they're still IBM PC's, and so are today's PC's. You could argue that we should be saying "IBM PC XT" or "IBM PC AT" instead of "IBM PC", but people don't watch to say all that.
    – Tim Locke
    Mar 21, 2018 at 13:26
  • 4
    An IBM AT used an ISA bus, just for starters. "Our PCs" most certainly do not. Trying to call "Our PCs" "an AT" is at a minimum very misleading. The reality is that architecture, branding and evolution got a lot more complicated than the single descendancy line that you suggest - very quickly - and using the "AT" label stopped being meaningful more than 20 years ago. Mar 21, 2018 at 19:06
  • At some point, after IBM got out of the PC business, Microsoft was licensing a "Made for Windows" trademark. Aug 24, 2021 at 12:19

4 Answers 4


First of all, the original IBM PC is capable of much more than is commonly believed, although that didn’t help with games back in the day of course (and the effects in that demo aren’t necessarily usable in a game anyway).

That’s not all that relevant to your actual question though. What is relevant IMO is the fact that game publishers were quite careful to document the real requirements of their games; see for example this 286+ game (Willy Beamish) or this 386+ game (The 7th Guest). This was important even in the early PC days: the very first PC itself existed in a variety of configurations, with different graphics adapters (commonly MDA, CGA, and then Hercules) and different amounts of memory. So stating that something required an “IBM PC (or compatible)” was never sufficient to know that any given computer would be able to run a specific piece of software...

Many would argue that PC gaming only came into its own with the advent of the 386, VGA and sound cards, none of which were present in the “official” IBM PC AT. Back in those days people certainly differentiated AT-class computers from 386-class computers, so saying that “PCs are in fact ATs”, while correct in some ways, would have been considered misleading in others even when ATs were still (somewhat) relevant.

It’s also worth considering that our PCs are now far, far removed from ATs, even if they’re still backwards-compatible with them (and even then, not for much longer). So “our PCs are in fact ATs” has become less and less true with time, and is now pretty much irrelevant.

  • And just how far we've come is illustrated by the fact that my Android tablet has a program on it that emulates a generic 286-based (or possibly even 386-based) system running MS-DOS with VGA or SVGA graphic capabilities - that is, it emulates an "AT" Mar 21, 2018 at 11:42
  • Even Zork was very specific: IBM PC, PCjr, XT, AT (PC DOS 2.0 or higher) 5 1/4 " double-sided disk, 64K) - i.pinimg.com/originals/98/65/7e/… Jun 18, 2019 at 8:46

Since the natural descendants of the PC are the XT then the AT, I believe that our PCs are in fact ATs, is that correct ?

No. If at all, they are PC-Standard Computers, as this group, lead by Intel andMicrosoft was the only standard body trying to define and standardize the advancements.

In terms of compatibility thay are no longer defined by some hardware API (*1), but software API's - thanks to an ever increasing CPU power enabeling the encapsulation of hardware to a great extend. The last of these common standards is the PC-2001 from, well, 2001. One reason for not adding any newer is that the 2001 standard did not only cover next to all common user related interfaces (and removal of obsolete), but also is seen as the point where the Legacy-Free-PC came to be.

It was based on the fact, that individual interfaces where no longer developed. Standards like PCIe and USB covering all I/O connections with new deviced only added thru one of these. All together hiden by multiple levels of API - just this time way too complex for average programmers to hack around :)

*1 -in fact, the PC was never realy to be defined by hardware, it was just badly written software bypassing abstraction layers that resulted in hardware dependance.

  • 2
    To be fair to early PC software, the abstraction layers available to them were often totally inadequate for what they needed to achieve.
    – Jules
    Mar 21, 2018 at 16:28
  • ... especially if they wanted to output 256 colours or use a sound card! Mar 21, 2018 at 17:46
  • @StephenKitt 256 Colours would be a VGA, and especialy here IBM offered an awsome software interface (similar, just less awesome) for EGA. And for Sound, there where great libraries to handle this in a less hardware dependant fashion.
    – Raffzahn
    Mar 21, 2018 at 18:03
  • 1
    @Raffzahn: Code using the supplied abstraction layer would take about a second to fill an 80x25 text screen, and many seconds to fill a 320x200 graphics screen. Software which bypassed that abstraction layer could fill an entire text screen in 1/30 of a second and a graphics screen in about 1/10 second. Using hardware directly allowed programs to be more than a full order of magnitude faster than would have been possible otherwise.
    – supercat
    Mar 21, 2018 at 22:35
  • 1
    @Raffzahn: Blame the designers of the architecture for not having provided a usable alternative in the days of the 8088 or even 80286. The design of the CGA meant that a routine which accepts a bunch of data to be displayed and shows it as quickly as possible in one operation can vastly outperform a function which has to be called separately for each character. There's no particular reason the BIOS couldn't have included such a function, but it didn't, and there's no way to get decent performance on the CGA without such a function.
    – supercat
    Mar 22, 2018 at 14:37

Today's "PCs" (as in Windows/Linux/... x86-64 box) have little in common with an IBM AT. There's still some compatibility on the software level but the hardware is vastly different.

  • Shugart FDD interface - gone
  • ST506/MFM HDD interface - gone
  • ISA bus - gone
  • DIN-style keyboard interface - gone (still somewhat common with the compatible PS/2 connector - but vanishing)
  • parallel printer port - gone
  • 25-pin serial port - gone
  • MDA/CGA/EGA video - gone

Even several successors of these vanished interfaces have gone:

  • EISA bus
  • VLB
  • PCI bus
  • parallel SCSI
  • 9-pin serial port (mostly)
  • VGA video (15-pin) (mostly)

Even the AT's BIOS that has incredibly long worked as a core for later BIOSes is increasingly replaced by UEFI.

Anything substantial left?

  • “Anything substantial left?” The x86-compatible CPU ;-). Quite a few modern systems still have 15-pin VGA output and 9-pin serial ports (which doesn’t detract from your main point of course). IMO the big, big change which is coming soon is the disappearance of CSM from UEFI, which means PCs will no longer be able to boot DOS... Mar 21, 2018 at 13:09
  • 2
    The "x86-compatible" CPU is a RISC CPU running an x86-compatible frontend and has extremely little in common with an 80286 - yes, it can still run (most of) the code but that's about it. You're correct, VGA HD-15 and serial DE-9 haven't quite vanished but they're about to.
    – Zac67
    Mar 21, 2018 at 17:44
  • 1
    Many of the interfaces you've mentioned age being gone are still there in a software backward compatible form. Most modern SATA controllers are IDE compatible which is ST-506 compatible. PCI-Express is software backwards compatible with PCI. Also if you want a 25-pin serial port or parallel port you can add one to just about any PC. You can even find modern motherboards with parallel ports right on the I/O backplane,.
    – user722
    Mar 21, 2018 at 18:27
  • I did mention the software vs hardware compatibility on top - yes, you can still refit legacy interfaces but they're gone from mainstream.
    – Zac67
    Mar 21, 2018 at 18:50
  • A lot of the things like FD, HDD, printer etc interfaces you are listing as "gone" were in the IBM versions of the AT only available on expansion cards - only on clones were they motherboard features. You can still get those interfaces on expansion cards or USB equivalents, so they are not "gone". Additionally, this isn't actually an answer to the question which was asked - it is not about modern systems, but rather about the systems for which these PC-ecosystem games were designed. Mar 24, 2018 at 17:32

A good metric of whether PCs still exist might be whether such a computer can run MS-DOS (or better, the successor OpenDOS which is actually still maintained). The answer for many of the computers we have in front of us right now would be yes. (There would be some hardware incompatibilities, as DOS drivers don't exist for a lot of hardware, but the machines would run, and some software could be loaded and executed.)

Other posters have clarified on the differences between the XT and the AT. I have always casually thought of the XT as 8088/8086-class machines and the AT as being 80286-class machines, and if that's a reasonable line of thinking (and I think it is), anything 80386-based and newer stopped being a PC AT, even if it would satisfactorily run AT software. (Bear in mind that the actual AT ran almost all XT software just fine, too.)

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .