Let's put this puppy together
The case that I picked out of Byte from the now-defunct XOR Corporation had a beefy power supply to feed its four eight-inch drive bays and S-100 slots. "Bays," however, is overstated - "holes" is more to the point, since you had to supply your own mounting brackets.
Power is supplied by a huge transformer that feeds equally huge capacitors
The S-100 bus in the XOR enclosure's "card cage" had a dozen slots - a normal amount at the time. I'd seen enclosures with more, and I'd seen enclosures with fewer. Even though I knew I didn't plan to use all the slots, I wanted to leave space for ventilation.
This S-100 bus card cage has room for a dozen cards
Installing cards was a snap - literally. You simply pushed them down into the slots at the bottom of the cage. First, though, you had to dig through opaque manuals and discover exactly how to set the cards' myriad jumpers and DIP switches - and if you don't know what a DIP switch is, consider yourself lucky. And young.
Interestingly, I discovered that the order of the cards affected how well the system ran - latencies and timing issues, I assume. Much of computer assembly in those days was done by the seat of one's pants.
At the base of the S-100 card cage are 100-pin slots
Although there were a number of reputable S-100 card companies in those day - meaning folks who actualy believed in such niceties as warranties - I was partial to Bill Godbout's CompuPro line, having used his machines at work. I could just as easily have gone for cards by Cromemco, California Computer Systems, or some other, but I stuck with what I knew.
The CompuPro CPU 8085/88 card - pick a processor
The CompuPro CPU 8085/88 (PDF) was a dual-processor card that contained a 16-bit AMD 8088 and an eight-bit Intel 8085. You could switch back and forth between the two in software, thus avoiding popping the case open and setting a jumper every time you wanted to switch.
AMD 8088 (left) and Intel 8085 - notice that each chip has just 40 pins
The 8088 was essentially an 8086 that could deal with the eight-bit S-100 bus. Its combination of 16-bit insides with an eight-bit approach to the outside world made it a good chip for the S-100 bus, offering a convenient way to use an eight-bit file system like CP/M 2.2 and still develop 16-bit code.
Next page: Card games
Everything old is new again...
So basically the floppy-emulator RAM card was my generation's version of the SSD. Except without that persistence without power thing...
This article made me nostalgic for Motorola's 6502 and then 680x0 family -- as a teenager earning spending money writing small business software on 6502 systems, interning writing assembler on 8086 embedded systems, and discovering the revolutionary 680x0 family at college, I was absolutely convinced that the elegant, orthogonal Motorola architecture would win out over the baroque Intel instruction set, or even its Zilog Z80 cousin... ah, the joys of youth...
As for the "what's the point?" whiners above... what exactly is your problem? Is there a shortage of space for articles? Was some other more interesting article pushed aside for this? Did you pay too much for an article you don't care about? Oh wait... this is free and online! So you're just making a fool of yourself complaining about it!
"wumpus , could not remember that name"
If you're running a Linux or BSD variant, try typing "wump" at a command prompt :-)
that's the name
wumpus , could not remember that name
But I do remember trying to decide on the s100 or a heath kit .
And the winner from Japan , boot leg Apple 2 mother BD...
thanks for memory lane down the Byte way
Windows 7 capable?
If Microsoft can trim down Windows 7 to run on this little beast, I'll buy that for a dollar.
Not really 16 bit8088 8086
Most S100 used 8080 originally and then later Z80 or 8085. Few used 8088/8086 as it wasn't really 16bit. 68000 was really the first S100 16Bit. I remember Cromemco UNIX like Cromix 5 years later on dual Z80/68000 "minicomputer".
the 8088 uses an 8085 8 bit data / 16 bit address multiplexed and 4 extra address bits on the 8088 for 1024K MAXIMUM. But the various peripherals and screen RAM was mapped between the 640K and 1024K. The 8086 had 16 bit data on external bus otherwise identical to 8088.
Additionally the 8086 and 8088 are not true 16bit CPUs but basically the 8085 with a few 16 bit instructions and only 64K memory blocks like the 8080/Z80/8085. Later a Z80 clone had segment page memory manager for 1024k and the 8bit Z80 in Amstrad PCW could access 512K.
So the 640K was really a limitation of the small segmented address space of the pseudo 16bit 8088 and 8086. The amount of contiguous RAM depended on if text, hercules, CGA, EGA or VGA to and extent. It was much later that TSRs (AKA dirvers) could be loaded into unused holes in the 1024k as early DOS didn't support it and early IBMs had 128K, 125k, 384k, 512K or 640K physical ram. Later 1M physical DRAM was used.
The first real 16bit IBM PC was the 286, but DOS used it as an 8086. Only UNIX flavours of OS used it as a flat 16M address space.
The PC and DOS held back PC computing nearly 10 years as it was really not much different to the CP/M and 8080/8085/Z80 8bit machines, just more standardised.