CREEPINGNET'S WORLD
GENERAL STRUCTURE OF THE IBM COMPATIBLE x86 PC
This page covers universal things that apply to pretty much all IBM Compatible x86 PC's from the original in 1981, to even some stuff still applies today with modern Core i-series powerhouses.  Also here, I'm going to dig deeper into the computer archetecture stuff more - like IRQ's, pinouts, and Memory Addresses, or BIOS settings and what they mean.

BOOT PROCESS - PC'S

  1. USER FLIPS POWER SWITCH - a  rush of electrons come in through the power port and kick the power supply awake, power supply inquires the motherboard as to whether it's ready for the power good signal yet
  2. MOTHERBOARD GIVES POWER GOOD SIGNAL BACK TO PSU - Motherboard sends a signal back to the motherboard telling it the PC is going to POST (Power On Self-Test)
  3. POST - This is when most people see the boot process taking place.  On most older PC's like we handle here, the first thing you'll see is a Memory Counter and maybe some branding and BIOS version/Copyright date messages.  Once all memory is accounted for, the PC silently checks all the other devices against the list in CMOS and either displays any relavent error messages and/or gives error codes using the internal speaker - or all is good to go, so the PC beeps once and then enters the bootstrap phase.
  4. BOOTSTRAP - System is okay to enter bootstrap, so, The boot devices are executed in order of the CMOS or in the standard order (Floppy, then Hard Disk, then auxilary devices).  Track 0 of the disk is read and the O/S begins it's loading process.

PROCESSOR CHIPS, TYPES, AND WHAT THEY ARE CAPABLE OF IN GENERAL + COPROCESSORS

The IBM Compatible PC turns 40 years old as a platform this August, that's 40 years of x86, IBM Compatible, Intel powered PC's. That also means generations upon generations of such hardware, with generations of evolution between them. Maybe you're new, and you don't know why we refer to these in cryptic numbers such as "8088" or "486", or maybe you wonder what someone means by "AT" or "XT". That's what this section deals with.

Generation 1 - 8086/8088/80186 - aka "PC/XT" Class - (1981-1990)
Gen 1 PC's start with the original IBM Personal Computer 5150 released in August of 1981, and ends with the last XT Clones still being built up into 1990, just as the 486 class was taking over. These machines were based on the Intel 8088 microprocessor by large, a budget version of the intel 8086, which was also used occasionally on certain makes and models. Very rare in this class is the intel 80186 which was used in some later XT workstations toward the mid-80's. In 1987, the original IBM PC and IBM XT were discontinued, and by 1991, nobody was making clones of them either.

This class is obviously the most limited. Early software was mostly limited to text mode business applications intended to be run on regular Green or Amber Monochrome displays. However, this era is also highly underestimated, and actually rather capable of a lot of things people would not think a 16-bit IBM Compatible could do. Early games were not even for MS-DOS, they came on bootable floppy disks that basically ran the game's bare kernel like a cartridge console game would from the PC's memory. A strong XT with 640 kilobytes of RAM and a decent sized Hard Disk of around 80 Megabytes or more is quite a formidable machine, even as a practical retro-user like myself, as most XT-class applications are VERY Tiny, some as little as 4-5 Kilobytes in size. And these applications can be very deceptively simple, looking like a boring command with a few switches while in typical use, but actually being able of doing great things in capable hands. Later games, starting around the time of the IBM PC JR. and Compaq Portable,. really started to show what this class was capable of. 16-color PR Jr/Tandy/EGA graphics, 3 voice GameBlaster or Tandy 1000 3-voice sound (derived form the PC Jr.), or even Adlib or SoundBlaster music was possible on an original 1st Generation IBM Compatible. It just was not a common option at the time because of how expensive these things were at the time. They cost as much as a regular PC today in their time, which equates to the cost of a small Economy car at the time. You could buy a Toyota Corolla brand new for about $6000, you could buy an IBM PC XT with all the trimmings and fixings for around the same cost.

What ended the 8088's reign on PC computing was not the next generation, but rather the generation after it, the i386 class, as they were fully 32-bit CPU's out of the box and were capable of running far more sophisticated software, and not crippled by some pre IBM PC incompatibility like the 286.

Generation 2 - 80286 - aka. "AT" Class - (1984-1992)

The second generation PC is based on the intel 80286 Microprocessor or "286" for short. The Intel 286 is a 16-bit CPU just like the 8088 on the PC/XT class machines was, but also contained a much faster 24-bit Memory Address bus. The intel 8088 was designed in 1978 as a "general purpose" CPU, and the 286 in 1982 for much the same reason - pre PC. AS such, the 286 had some compatibility issues with it's new "Protected Mode" - an operating mode that allowed the CPU to access up to 16 Megabytes of RAM - a capacity we would not see on the norm until the late 486, early Pentium era in the mid 1990's, and allow the CPU to Pre-Emptively multitask between multiple programs, such as on a Unix workstation (the spiritual ancestor of Linux by function). The problem with this though, was that once in the new enhanced Protected Mode the CPU needed to be reset (rebooted) so the computer could return to "Real Mode" used by DOS. This caused problems with software development and garnered many headaches. The problem with the 286 was that there was no problem, people just did not understand what IBM was intending it for when they released the IBM Personal Computer AT 5170 in 1984 - which was as a high end business workstation intended to cross-interface with mainframes and run Xenix, Microsoft's own Unix clone at the time.

Despite all the hoopalah about the 286 being "Braindead", it truly is quite a capable CPU, and by the end of it's lifecycle, programmers had learned tricks and techniques to work around it's inherent Pre-PC Design limitations, making it quite a capable machine. During the life of the 286, "EMS" or Expanded Memory Systems became a new thing, allowing one to use more tha 640K RAM in DOS, allowing for bigger applications. Most people joke about the 286 being incapable, but the truth is it can do a lot of things it's younger 386 and 486 siblings can do, it just can't do them as fast or as smoothly in a lot of cases. So much so the Intel 286 is the baseline CPU for MANY great golden-era DOS games to run smoothly such as Ultima VI: The False Prophet, or The Secret of Monkey Island - which while those will run on an XT, it won't be fun unless it's a 286 or better CPU. All in all, the 286 was mostly used as a "Fast XT" as even at the same clock speed it was much faster. Another big deal of the time was the "Zero Wait State" operation - something my GEM 286 has - which mean that the computer did not need to pause to let other parts of the system catch up. Also, the 286 introduced the 16-bit ISA expansion bus to PC's which was the standard from 1984 all the way up until 1995 when the Pentium finally took over.

Generation 3 - 80386 - aka. "386" Class - (1986-1995)

The 3rd generation is the beginning of IBM Slipping out of their guilded position as a PC maker as the Texas company Compaq beat IBM by a full 9 months getting their first 386 product to market - the Compaq Deskpro 386. The Compaq Deskpro 386 was easily one of the fastest, if not the fastest commercially availible PC at the time, and thusly very expensive, hence why the 386 needed almost 4-5 years before it became commonplace on desktops both in the workplace and in the upper-middle-class or better home. THe intel 386 is a fully 32-bit CPU - meaning it can move 32-bits of data at a time. It also now had a new Protected Mode that was hot-swappable between Real Mode, and was utilized by Microsoft and IBM for their OS/2 Project, and later Microsoft WIndows 386 and later - to bring the Graphical User interface to the desktop, signifying the end of the DOS era for the mainstream

With this new power, more elaborate software could be made. Sure the old 80x25 column "Killer Apps" such as DBASE III, WordPErfect 5.1, or Lotus 1-2-3, were used in 80x25 column text mode, but this meant that you could process bigger Spreadsheet data, make bigger documents, or create more complex databases, and the machine would not need a few hours to sit around and render everything, or choke on the data it was being expected to deal with. Hand-in-hand, the 386 obviously allowed for very complex games, more complex than most home consoles at the time. The 386 was the true processor generation that was a legend in DOS Gaming, a acclaim given a lot to the 486 these days. Ultima 6 and 7 were both at their prime on 386-based hardware, as was the glut of new 256 color VGA Games coming out at the time from Sierra, Lucas Arts, E.A. Games, and Maxis. The 386 was the original i7, the original powerhouse PC for business people and hardcore gamers.

Generation 4 - 80486 - aka. "486" Class - (1989-1997, 2007)

The intel 486 came out in 1989 and is likely the most interesting PC Generation to get into because it's literally the PC's "Adolescence". AT the start of the 486 generation in 89', the 486 was a high end powerhouse PC like a Core i9 system is now, and was mostly used by businesspeople for large spreadsheets or by engineers for things like Computer Aided Design and complex scientific calculations in STEM fields. But at that time, it was still basiclaly an x86 IBM Compatible computer running on DOS. BUT, as the 486 era went on, DOS gave way to Microsoft Windows, and especially after the Internet went mainstream and Windows 95 set a lot of standards still in place now, the 486 starts to look more and more like a modern system that existed 20 years ago.

The intel 486 is an incredible machine, and even in some ways, quite usable as a daily driver even today almost 30 years later, which is insane when you think about it. Many modern tropes and standards in the computer industry started with the 486. User upgradable CPU in Zero Insertion Force sockets, user upgradable memory that did not require understanding of the internals, just the capacity. CD-ROMS, the Internet, the modern standard GUI design, Specialized High-speed Bus ports (VESA Local Bus, PCI, EISA), high speed secondary Caches, graphics modes over 640x480 at 256 colors (SVGA, XGA), high speed hard drives in capacities over 1GB, and CPU's over 100MHz in speed were all things that came about in the 486 Era. The Intel 486 went from yet another glorified XT running at ridiculous speed, into a 32-bit Graphic User Workstation surfing the internet, sucking data off USB Flash drives, and shoveling data over a TCP/IP Ethernet network at 100mbps shows just how amazing this era was. The 486 era started off looking no different from the 386 era, VGA DOS Games with internal speaker, Adlib, or SOundBlaster sound, such as Wolfenstein 3D, Eye of the Beholder, or Monkey Island 2, or Ultima VII....but by 1995, was starting to look more and more like modern gaming with Doom, Quake, Postal, Grand Theft Auto, Red Alert, The 7th Guest, or Diablo with full motion video, CD Quality (or better) sound, digital music soundtracks like a store bought CD, and graphics modes of 800x600 at 16-bit color or better in some cases. In just 5 short years, the 486 took the PC INdustry from "yet another glorified DOS-based XT" to "Ultimedia POwerhouse Workstation".

Generation 5 - Pentium Class Hardware - (1993-1999)

The Pentium, or 586 as it's almost never called, had a tumultuous beginning, like that time in your early 20's where you move into an apartment by yourself for the first time and have to learn how to manage finances while still saving enough to have fun and creating an emergency fund - and screwing it all up several times. The Pentium first came out in 1993 in 60Mhz and 66MHz incarnations, that to some tech types were described as being somethhing more like a Dual Core 486 sharing the same math co-processor and write-back level 1 Cache. These early Pentium CPU suffered the same problems as the Pentium D would almost 10 years later - they consumed a TON of electricity, sometimes to the point the then high-output 250 watt AT POwer SUpplies of the time could not handle the requirements. They generated a TON of heat meaning new strategies in thermal management, especially compared to the 486 DX2-66 in which could manage to survive without a heatsink. Also there was an issue with certain calculations because of a floating point math bug in the CPUs. Intel took the next 2 years - just in time for Windows 95, to get it right.

In 1995, that's the true beginning of the Pentium processor's rise to dominance in the market for the next 8-10 years would begin as a part of a package that seems to happen once or twice a decade with computer hardware and software to this day. The internet just became mainstream as the new "to get" technology, Windows 95 was the latest version of a Microsoft O/S for the machine and set the standard for all WIndows releases going forward, and the Intel Pentium processor was the package deal with the whole package. So there you have it - the first generation 100% like today's generation, just older, slower, and less-as-capable. 1995 has to be one of the most important years in PC history.

In the case of a Pentium Processor, many of present-day CPU capabilities exist and it has the widest pool of sofware availible to it save for maybe the 80486 generation. Pentium systems were powerful enough peoploe were still using them well into the 2000's as their primary computer in the home, and even some businesses who had specialized software were still running these as newer platforms, expecially from the Pentium D of 2005 and onward which were 64-bit, started to gain a foothold for daily use and some specialized applications would not run on anything newer than a 586 system due to too much speeed or differences in internal wiring.

The Pentium era, in gaming and software in particular, is when we started to see WYSIWYG (What You See is What You Get) productivity software such as HTML Editors, Word Processors (Microsoft Word 97' for example), Spreadsheets (Excel), Frontpage, DiDaPro, Mozilla Composer, or PowerPoint (heck, the whole dang Microsoft Office Suite became a big thing at this point in business). In gaming, sure, your 486 could run Quake, Doom, or Diablo, but the Pentium and the PCI Bus it had gave it capabilities to give a experience with some of those newer games it could run on par with a modern system in graphics and sound. This was also where GPU started to become more and more important as more and more of the games graphical handling was handed off to things like the TNT Riva, S3 Trio, ATI Rage, or even some of the earliest NVIDIA GPUs.


BIOS, CMOS, & SETTINGS


Starting with the IBM PC AT on up, one can save the configuration of the computer's basic input output system to a CMOS chip on the motherboard via either a setup diskette (like the original IBM AT, or the Compaq Deskpro/Portable products), or a special key combination at boot time (ie F1, F2, CTRL+ALT+ESC, F10, F8, CTRL+S, ALT+S...there's a bunch of different ones depending on your motherboard/BIOS/Computer manufacturer). Since this is pretty long, I'll just link to this page here

Not all of these BIOS Settings will apply to your system, some only existed on earlier systems, some on later systems. THere may also be some more "Technical" explainations on this page for things, but you can always ignore those and just see what he reccommends as being enabled or disabled. IT's a good place to start learning how to configure your own system


IBM PC/XT DIP SWITCH SETTINGS & ARRANGEMENT

By contrast, older IBM PC and XT style computers utilized a series of DIP Switches on the motherboard to set the same settings we used the old CMOS/BIOS settings above to do back in the old days. The best tech reference for this is minuszerodegrees.net which is practically a compendium on IBM specific hardware such as the original XT and AT. Since this was cloned, a lot of "Turbo XT" motherboards also utilize the same dipswitch settings.

Tandy 1000 owners would want to go to TV Dog's Tandy 1000 Archive for information on their dip switch settings since the Tandy 1000 is kind of it's own special beast and has some very different jumper and Dip Switch settings for setting system options.

IRQS - or Interrupt ReQuest Lines

Hardware Interrupt Requests - or IRQs for Short - are sort of like the coattails of the hardware in the computer.  Basically, the CPU tugs on one to get the attention of a hardware device in most cases.  IBM Compatibles have a pretty standard IRQ to Device Mapping that is utilized.  Today it's a little less important to know because of Plug And Play being solid and reliable and because IRQ steering allows for an insane number of devices on the computer already.  My table below pretty much applies to all PC's, but specifically these two tables are aimed at legacy PC's from the original IBM PC 5150 through early-mid era Pentium systems running Windows 3.1x or Windows 95.  Only one device can utilize an interrupt, or otherwise, an interrupt conflict occurs, which can cause devices either not to start, work, or cause the system to hang.

 IRQ# TYPICAL DEVICE OTHER DEVICES
 0 System Timer  This cannot be changed
 1 Keyboard (controller)  This cannot be changed
 2 Open Port  Usually this is used for early SoundBlaster Cards, also has been seen to be used for PC Jr. Sound on IBM PC Jrs and Tandy 1000 series PCs.
 3 Serial Port (COM 2)  Sometimes this is disabled to free up for other dev
 4 Serial Port (COM 1)  Usually used for Serial Mouse or External Modem
 5 Hard Disk Controller  MFM/RLL/ESDI/SCSI Controllers
 6 Floppy Disk Controller  Cannot be Changed
 7 Parallel Port (LPT 1)   Sometimes this is disabled to free up for another device.

IBM PC AT (80286) and Later PC's
When the IBM PC AT was released, the 8259 Interrupt controller chip on the motherboard was doubled and typically cascaded from IRQ 2 to IRQ 9 - this is the link between the two. 
 IRQ# TYPICAL DEVICE OTHER DETAILS
0 System Timer  This Cannot be Changed
1 Keyboard Controller  This cannot be changed
2 Cascade from IRQ's 9-15  This is the connection to the second 8259 Interrupt Controller chip and takes signals from IRQ's 9-15.
3 Serial Port (COM2)  Sometimes disabled to free up for another device such as a sound card or network card
4 Serial Port (COM1)  Sometimes disabled to free up for another device
5 Sound Card or aux LPT  Most often used for SoundBlaster Cards, sometimes used for network cards
6 Floppy Disk Controller  Same as the XT, never changes
7 First Parallel Port (LPT1)  Sometimes disabled to free up for another device
8 Real Time Clock (Date/Time)  This is the part of the CMOS/BIOS for Date/Time
9 Usually Free or Sound Card  Usually used for a Sound Card as an alternate IRQ if the others are taken up, also used for network cards.
10 Usually Free or Sound Card  Usually used for a Sound Card as an alternate IRQ if the others are taken up, also used for network cards.
11 Usually Free or Sound Card  Usually used for a Sound Card as an alternate IRQ if the others are taken up, also used for network cards.
12 PS/2 Mouse Port This never changes.  Might be a free IRQ on some regular AT class machines or machines without PS/2 port(s).
13 Math Co-Processor Cannot be changed, might be free if no CoProcessor is present.  I suggest leaving this one alone.
14 Hard Disk Controller (Pri) Primary Channel of IDE controller. Free if system is Floppy Only (very rare for AT+ class machines)
15 Hard Disk Controller (2nd) Secondarly Channel of IDE Controller or free if single channel IDE present.

FLOPPY DISKETTE DRIVE TECHNOLOGY

IBM Compatible PC's in general utilize at least one of four types of floppy drive:

  • 360K DSDD, 5.25"
  • 1.2M HD 5.25"
  • 720K DSDD 3.5"
  • 1.44M HD 3.5"

360K came first and actually replaced a very early Single Sided 160K and intermediate 320K Floppy drive on the original IBM PC.  The 360K diskette became a standard first.  So all 8088 XT class machines are compatible with this standard.  The diskettes are told apart by an extra "Hub Ring" added to the center of the disk where the motor spins it from, as well as an index hole for motor calibration.   These disks are also actually floppy.  The drive, you will have to look up the model number to know if it's an actual 360K Drive or something else.  DSDD stands for "Double Sided, Double Density".  With some non-IBM Vintage computers, you had to take out the disk and flip it over to use the other side - like the original 160K drives.

1.2M HD 5.25" appears to have come afterward starting with the IBM PC AT in 1984.  Despite being able to READ 360K DSDD, it cannot write those disks because the tracks containing the data will be too thin for a 360K drive to reliably read.  So it either gets read out corrupted or won't read at all in a naitive drive.  This is why a lot of "AT" Class machines had 2 drives in different colors or had stickers on them telling you which drive was 360K and which was 1.2M.

720K DSDD 3.5" is kind of an interim standard of the mid 1980's that never really caught on in such a large amount, because by the time these disks were starting to be somewhat commonplace, the ubiqutous 1.44MB 3.5" Floppy Disk had become standard, and ulike the 5.25" drives, it could read and write 720K disks as well as 1.44MB 3.5" diskettes.  This meant backwards compatibility and made owning a dedicated 720K drive rather un-necessary.  How to tell a 720K Diskette from a 1.44MB diskette is by the fact there is only ONE hole on the top of the disk- which has a slider to "write protect" the disk.

The 1.44MB 3.5" disk looks almost identical to the 720K disk with the exception of a second hole that tells the floppy drive that this is a 1.44MB disk and to act accordingly.  1.44MB is still the easiest format of the lot to get.

Currently, as a future-proofing measure as Floppies are not in mass production that much anymore, companies such as GoTek are offering "Floppy Emulators" - which are devices that attach to the Floppy controller and behave as a floppy drive, while reading data off of floppy image files, partitions, or directories on a USB Key.  You use a interface on the front of the drive to select which disk image off the USB Stick the GoTek Emulator is to use.

HARD DISK DRIVE TECHNOLOGY. CF CARDS, IDE, SOLID STATE DRIVES IN THE VINTAGE PC CONTEXT

The first IBM Compatible with a hard disk from the factory was the IBM PC XT released in March 1983.  The IBM PC XT featured a 10MB full height MFM encoded ST-506/412 standard hard disk drive.  These drives consumed a lot of power, had slow seek/read/write times, were noisy, and also quite finicky compared to modern drives.

ST-506/412 - aka. MFM/RLL Drives

The ST-506/412 standard was started by Seagate, who made the ST-506 5MB and ST412 10MB HDD, the latter of which was availible on the IBM XT.  These drives were 5.25" full height devices that were loud, finicky, and power-guzzling.  They also were VERY expensive.  To put a ST-255 in a Tandy 1000 in 1985 would have cost you $4000.00 per Radio Shack's old print catalogs.  Popular drives in this format are mostly Seagate models including the ST-412 (10MB), ST-225 (20MB), ST-251 (40MB), ST-238 (40MB RLL).

MFM/RLL refers to the encoding method used to store data on the HDD.  MFM is "Modified Frequency Modulation" - which is also the method used for storing data on Floppies.  RLL is "Run Length Limited" and allowed for tighter packing on a hard disk leading to the ability to store more data.

These drives used 2 cables - one for data - which was the smaller cable, and there was one per drive, and then the actual "control" cable which was a regular floppy cable, except usually without the twist.  The final drive on the chain would get a "terminating resistor" installed, and there was an extra step to formatting and partitioning the drive called a "Low Level Format" - which meant you usually typed some kind of special command into Debug from DOS (usually G=800:C if you were using a Western Digital controller) to invoke the low-level format utility.  Which would low-level format the drive, and mark all bad sectors.

ESDI - The EGA of Hard Disks

ESDI, or Enhanced Small Drive Interface, was another standard that turned up between ST-412 and IDE.  It stood for "Enhanced Small Drive Interface" and is quite rare.  It seems IBM PS/2 used this for awhile before moving to IDE in the early 90's, however, despite having over 60 different systems since 2001 I've never encountered any actual ESDI drive.  They are like EGA in that they are not that common to the point collectors don't believe it's a real standard in some circles.

SCSI - Small Systems Serial Interface....

SCSI was another interim standard that never really took off on x86 PC Desktops, however it did take off on Apple Macintosh products.  It was mostly used for Tape Drives, early CD-ROMs, Magneto Optical Drives, Bernoulli DRives, Zip Drives, Jaz Drives, stuff mostly relegated to the Server Closet or high end workstations that required some form of high capacity, ultra high speed data storage.

SCSI could allow up to EIGHT devices (including Controller Card) on the same controller and have all of them be bootable.  This was what made it attractive to people like SysOps, Systems Administrators, and other "behind the curtain" I.T. stuff of the time.

IDE - Integrated Drive Electronics, aka PATA, aka ATA

IDE was introduced sometime in the late 1980's and became a standard and stayed until well into the late 2000's.  This is what we usually use with Vintage computers today as it's widely availible, plenty of adapters and older drives still work with it - especially on 386es and up or XT/286 with an XT-IDE card in it.  The most common setup today for vintage DOS machines  is to use an IDE to CF-CARD reader with the cards swappable through the back expansion slot area allowing for several tiny, storable Hard Drives you can use.

The earliest IDE PC's were limited to 528MB of hard disk space at most by the BIOS, then this limit was broken in the mid 1990's allowing for 2GB in most BIOS, later 8GB, and then 4TB. 

Older systems can see bigger drives using a piece of software called a DDO or "Dynamic Drive Overlay" which is written to the boot sector to translate the hard disk size to the computer during boot time allowing for use of the drive's full capacity (at the mercy then of only O/S capacity limits ie FAT-16 or FAT-12 capacity limits in DOS). 

OPTICAL DISK DRIVES (CD/DVD)

CD-ROM Drives first started to appear in 1987 at a premium, and never really took off until the early 1990's when 486-based, sound-card equipped, "Multimedia" PC's became all the rage.  Early CD-ROM Drives, such as the much sought after "caddie" loaders of the early 1990's, utilized strange, proprietary interfaces, or SCSI at best.  The first ATA drives started to appear around 1994-1995 and were slow 1x and 2x units that are usually broken.

Usually today, we put in whatever we can get ahold of, especially if it has a Beige or white Bezel on it.  This can mean anything from an old 2X CD-ROM from 1994, to a full blown 48x DVD-RW with Lightscribe from 2008.  Obviously, faster drives will yield better read/write performance.

GRAPHICS TYPES, CARDS, MODES, CHIPSETS, AND PORTS

MDA - MONOCHROME DISPLAY ADAPTER
The first standard released for PC's was Monochrome.  Monochrome is basically black & white with
a limited shade of whatever color the phosphors coating the inside of the monitor tube are.  A lot of people refer to Monochrome as "Green Screen" due to the commonality of "Green Phosphor" monochrome CRT back in the day.  But they also came in Amber and basic Black and White as well.

CGA - COLOR GRAPHICS ADAPTER
CGA was released in 1982 and did not have a dedicated monitor from IBM until 1983 (the IBM 5153). CGA typically displays 16-colors in text mode, 4 colors out of the 16 color palette in graphics mode.  Maximum resolution is 640x200 in black and white, or 320x200 with 4 colors out of the 64-color palette.  This was a common graphics standard on IBM PC XT class systems and some early 286 "AT" class systems.  It fell out of favor after VGA took off, though EGA and PC Jr. Video kind of carried the torch for awhile.

TGA - TANDY GRAPHICS ADAPTER/PC Jr. Video
TGA is a term given to the expanded CGA that the IBM PC Jr. and Tandy 1000 series had.  This expanded CGA allowed for a 160 pixel mode with 16-colors, and a 320x200 pixel mode with 16 colors similar to CGA.  It never really was availible on anything other than a Tandy 1000 or PC Jr. so it's sort of a "Sub-STandard" of sorts.

HERCULES - GRAPHICS IN MONOCHROME
Hercules as a short lived "Budget" video standard for PC's that allowed one to play games that utilized graphics modes such as Microsoft Flight Simulator or Sim City - on a Monochrome Display such as the IBM 5151 that shipped with most PC's and XT's from the factory.  It seems it fully fell out of favor when VGA become the fully ubiqutous standard circa 1988.

PGA - PROFESSIONAL GRAPHICS ADAPTER
IBM Offered a 640x480 256 color graphics card a couple years before VGA came out, aimed at CAD designers, and archetects, called the "Professional Graphics Adpater".  It was a 2 slot card like a modern video card that utilized a special monitor,  They are extremely rare and hard to find, hence expensive, but I'm listing it just in case you come across one in your travels.

EGA - ENHANCED GRAPHICS ADAPTER
EGA Graphics came next, offering similar modes to the Tandy 1000/PC Jr. Video, but with extra speed, and a 64 color palette.  Most often EGA used a 16 color pallet in the standard 16 colors of CGA text mode, but in graphics mode as well.  It also was not interlaced on text mode which made text crisp and clear compared to CGA.  Often EGA monitors would offer options to change the phosphor color to match a monochrome monitor as a gimmick. 

MCGA - MULTI-COLOR GRAPHICS ARRAY
MCGA was a short lived "neutered" version of VGA, most notably offered on the original IBM PS/2 Model 30 and MOdel 25 in 1987.  You were still limited to 16 color 640x200 pixel interlaced video for high resolution with the most color depth to that point like EGA, but you had the 320x200 256 color graphics mode offered by VGA in it - which was the most used graphics mode for DOS games starting in 1988 onward.

VGA - VIDEO GRAPHICS ARRAY
VGA came out with the higher end PS/2's in 1987 and very quickly became the widespread standard for PC's by 1989.  VGA offered the popular 320x200 256 color mode MCGA also had, but also had a high resolution 640x480 16-color mode that became the baseline resolution for Microsoft Windows starting with Windows 3.1 in 1992.

SVGA - SUPER VIDEO GRAPHICS ARRAY
SVGA is basically an enhancement on the VGA graphics system, but adding additional memory and occasionally and a special RAM DAC (Random Access Memory Digital to Analog Coinverter) to allow for high speed, high resolution (640x480+), high color (16-bit color or higher - or 24,000+ colors), graphics - later with 3-D enhancement either through an add-in card (3D Blaster) or as a part of the video card it'self (TNT Riva, 3DFX VooDoo, NVIDIA GeForce, ATI Rage....etc...).  It used the same connectors as VGA and managed to stick around well into the 2000's before other connections and higher resolutions beyond 1280x1024 at 32-bit color (32,000 colors) became a standard.

Monitors Used with Standards

Hercules, Monochrome
Hercules and Monochrome typically use a dedicated 9-pin TTL (Transistor/Transistor Logic) Monochrome Monitor of some kind, like an IBM 5151 or Compaq DSM.  These were best remembered in being all green because the color is based on the phosphor coating inside the picture tube and "Green Screen" was the most popular, with "Amber" being the second most popular phosphor color.  Later MultiScan Monitors can work as well, such as the NEC MultiSync series or some Diamnond Scan Mitsubishi monitors, and often these will provide the ability to switch between Amber, Green, or Black & White Monochrome screen emulation at slightly degraded quality.

CGA
CGA typically uses a RGB Color monitor or Television Set, either composite or using the 9-pin D-Sub connector.  CGA Monochrome is capable of using visual artifacts from the Composite signal to give more colors to colorful applications and games.  These monitors are well known for their terrible low-dot pitch so they are not exactly that crisp for text regardless of input chosen.  Examples include the IBM 5153 and Tandy CM-5.  SOme early Multisync monitors and all EGA monitors can handle CGA Signals with better picture quality.

EGA
EGA uses it's own higher-dot-pitch RGB monitors that utilize the 9-pin connector, some may have composite but most don't.  EGA can be used with CGA monitors but you have to be VERY careful what you run on them because higher resolutions can fry and damage the monitor or video card (or both) - so best left to experts.  EGA was not a very common standard so monitors are rare as the cards themselves.  Often these were paired with MultiSync monitors such as the NEC MultiSync and Mitsubishi DiamondScan monitors.

VGA, MCGA, & SVGA
VGA differs from the others in that it's an "Analog" standard, and also has higher resolution than any of the aforementioned that mostly top out at 640x200 pixels, when VGA can do 640x480 or higher.  So these required new special monitors capable of higher resolutions, and higher dot pitch.  They also utilize a full 15 pin connector that is still commonplace today on LCD's.  The earliest monitors were designed to only handle VGA or low-resolution SVGA modes (640x480 or 800x600 at 2 color mono).

MultiSync Monitors/MultiStandard Monitors
There are some "wildcard" monitors out there, probably the best known being Mitsubishi DiamondScan, NEC MultiSync, and Samsung SyncMaster.  These Monitors were capable of all video standards on their old examples, and with adapters could even be connected to other computer platforms such as Amiga, Commdore 64, and Tandy TRS-80, as well as newer standards such as VGA or SVGA.  A fine example is the NEC MultiSync JC-1402-HWA above, a 14" Multisync monitor capable of TTL Mono, Hercules, CGA, EGA, TGA, VGA, and SVGA being as it tops out officially at 800x600, and unofficially at 1024x768 (not bad for 1988).  It also can be tailored to work with other color modes and other systems such as the Amiga, Commodore 64, Tandy CoCo, or TI-99 with adapters and manual switch settings in back.


SOUND SUBSYSTEMS & CARDS - A CANDY STORE OF OPTIONS

Vintage Options

Internal Speaker (1981)
The "PIT" or "Programmable Interrupt Timer" chip, also known as the intel 8253, is the part responsible for the classic PC "beeper"/"tweeker"/"tweeter"/"blipper"/"bleeper" speaker that most people think of when they think "vintage PC".  It's a single channel, square wave audio device inititally intended to just give status-beeps to the user for feedback.  However, of course, like anything with a CPU and RAM, they figured out how to get games onto it, making this the first official sound device in IBM Compatible x86 PC's of the 1980's.  It still exists on modern PC's today as it's still used for giving BIOS error messages, particularly if the screen can't come up and that's the only form of feedback you can get.

PC Jr/Tandy 3-Voice (1983)
This was initially a TI SN76489 and similar audio chip used often in arcade games of the time that was added to the IBM PC Jr. - a budget (if you call $1000 budget) desktop PC aimed at the home user market in 1983.  When the PC Jr came out, Tandy Decided to merge it with the regular ISA based PC Architecture and created the 1000 Series - possibly the most popular IBM x86 Compatible of the 1980's.  On the Tandy it was marketed as "Tandy 3-voice" audio for it's three basic Square Wave channels.  On later 1000 models starting with the TL series - an 8-bit Digital Audio Converter was added allowing early 8-bit Digital Audio a full 3 years before the Creative Labs SoundBlaster came out.

Creative Labs GameBlaster (1985)
Creative's first card.  This was very similar to the Tandy 3-Voice, but not as widely distributed and not as widely supported as the 3-voice.  Early SCUMM, AGI, and SCI titles plus Ultima VI were among some of the games that supported this card. 

COVOX SoundThing/Disney Sound Source (1985)
These were Parallel Port based audio devices that looked a bit like a desktop intercom device.

AdLib Music Synthesizer Card (1987)
This is what most people think of when they think of late 80's/early 90's "SoundBlaster" game music, because Creative Labs adopted a compatible circuit for their SoundBlaster cards in 1989.

MPU-401/MIDI (unknown)

GRAVIS Ultrasound (unknown)

Windows Sound System  (1992)
Windows Sound System was an attempt by Microsoft Corporation to create a universal sound standard for Multimedia based around their Microsoft Windows 3.x GUI application (it was not considered a real O/S until Windows 95).  Aztec and Crystal Semiconductor (CS4231) were two major makers of compatible chips with this system.   Some of these have OPL, like the original Microsoft Card, while others, especially a lot of laptops with the Crystal Chip, don't (Crystal Business Audio).  DOS Compatibility was iffy at best with only certain titles supporting it such as The 7th Guest, Under a Killing Moon, most Sierra late SCI titles like Freddy Pharkas and Hoyale Card Games, but others like Monkey Island and Ultima did not.  SoundBlaster Emulation could be had through Microsoft Windows, often requiring a special driver set including EMM386.EXE and WSSXLAT.EXE to make it work.

Creative Labs SoundBlaster & Compatibles (1989)
In 1989, Creative Labs created the default standard for audio cards from that point going forward, until the 2000's at least.  Creative Labs Soundblaster merged 8 and later 16 or 32-bit DAC with a Adlib Card.  The earliest cards were 8-bit only, and up through the CT-1600, SoundBlaster Pro 2.0, even could have GameBlaster compatibility.  The ubiqutious SoundBlaster 16 series of cards were among the most popular and biggest thing through the 486 era, and after that came the AWE32 and AWE64 cards + PCI Variants of the 16.  These later cards were capable of as much as multiplexed digital audio (meaning early DAW functions) and using MPU-401 and SoundFonts, making the late models of these cards, especially in ISA Format, particularly sought after, and at times, rather expensive these days.

Creative's chipsets were not the only ones, it was so popular many compatible chipsets came out from many companies such as Crystal Semiconductor, Aztec, ESS, among others.   SoundBlaster  and compatible cards have the widest compatibility of the entire lot in this list other than PC Speaker, because they were THE go to thing when they came out.

Modern Options

Various DIY-er electronics engineers have started both reverse engineering the rarer sound cards and also coming up with modernized solutions to getting decent audio out of a Pentium or older PC with an ISA Bus.  I have listed, detailed, and provided links for some of these below. These links came from this VOGONS Thread. I''m only going to name SOME of these, plus I'm adding one I'm watching closely. I may also be adding a project of my own as well.

NETWORK - THE PHYSICAL LAYER - WHAT TO FOCUS ON, WHAT NOT TO

For this section, I'm ignoring ALL Legacy technologies because there's really no point to them unless you enjoy torturing yourself.  ALL vintage x86 IBM compatibles from the first PC all the way to the most recent PC, is capable of connecting to a standard TCP/IP network over Ethernet.  All you need is a network card, Ethernet cables, and either a packet driver or an NDIS network driver in most cases for the operating system of which you are using (and possibly any network client it may require to have network functionality added, if not already a part of the O/S).

Ethernet cards can come in 10mbps ISA cards, 8 and 16-bit, some 16-bit cards have an "8-bit" mode even allowing them to be used in an XT without issues.  They also can come in a 10/100 variant in PCI, maybe even some Gigabit cards work with Pentium models.  Lastly,  PCMCIA cards, both 10mbps Ethernet cards in PCMCIA PC Card type, 10/100 in Cardbus, 802.11/b/g wireless at 11mbps in PCMCIA Type II PC Card, or Cardbus 802.11/b/g in 54mbps.

The only caveats you should have with networking a vintage PC comes from software related limitations, ie nobody has written a WPA method to connect DOS to WiFi yet, nor has anyone come up with a way to deal with TLS without slowing Windows to a crawl either when using the internet.

ISA NIC Cards are extremely common and can be bought for a song.  However, later PCI systems can get 10/100mbps cards that also have NDIS, ODI, and Packat Drivers for legacy operating systems like DOS, Windows For Workgroups, and Windows 9x.    These are a prime choice and paired up with things like the networking subsystems of Windows, or with Mike Brutman's mTCP suite, it can make copying files between legacy and modern systems using regular networking or services such as FTP a real breeze.  These should all work with standard CAT-5 and CAT-6 ethernet cables and work properly with all pre-existing TCP/IP enabled network hardware with DHCP.

WiFi is also possible but it's a little more tricky.  Generally, there are two cards to lookout for for pre-95 systems like Windows for Workgroups and DOS - these are the CIsco Aironet 340/350 cards, and the Orinoco based cards such as the Lucent WaveLan Silver.  These will work in DOS, Windows 3.1x, and Windows 9x, however, there are a few caveats - first off, you don't have WPA on any of them (except in Windows 98 SE where you can change WiFi Management to a 3rd party program and then make use of WPA-PSK),  So you may need to create a special access point for these devices on your LAN, or you can lock down your Cell Phone's hotspot really good and use that.

STANDARD PERIPHERALS - MINUS MONITORS

KEYBOARDS
PC Keyboards come in 3 standard varieties, XT, PS/2, and AT - with a middle ground subset of XT/AT Switchable keyboards from the late 1980's and early 1990's.

XT
Typically a Large DIN-5 Plug, 85 or less Keys, only 10 Function Keys (F1-F10), more often than not clicky, and quite heavy.  The best known keyboards were the IBM "Model F" distributed with the IBM PC 5150 and PC XT 5160, the COmpaq Portable Keyboard with it's capacitive foam contacts that have a tendency to rot (get in touch with TexElec for that).  While the plug is identical to one found on AT keyboards, the wiring is different so never plug a XT keyboard in to a 286 or higher computer without some kind of adapter box between it or you could fry the Keyboard or Keyboard Controller.

AT & AT/XT SWITCHABLE
With the IBM PC AT introduced in 1984, a new keyboard and new pinout started, the "AT" Keyboard pinout.  Earlier keyboards are usually based of Buckling Spring like IBM's ubiqutous "Model M", or Northgate's OmniKey series which used Alps Keyswitches, or Cherry Keyswitches in some other brands and models.  Later models toward the end of the 80's into the 1990's, aimed at budget conscious users, were based on modern Rubber Dome technology and far less desireable among keyboard and PC collectors.   Toward the latter end of the 80's

PS/2

OTHER PERIPHERALS YOU MIGHT WANT/NEED AND A BIT ABOUT THEM