I was lucky to own a programmable HP-25 calculator in 1975.
See
http://en.wikipedia.org/wiki/HP-25
However, the first real computer I owned was a Motorola D2 Kit in 1976.
It was an evaluation board that Motorola had introduced for its 6800 microprocessor (
http://en.wikipedia.org/wiki/Motorola_6800)
and came with 128 bytes of static RAM (yes, that's bytes) and 1KB of EPROM.
The CPU could run at 1MHz but from memory was clocked at 512KHz.
It came in a large folder along with documentation :-
http://allardschaap.woelmuis.nl/comp...la/6800map.jpg
You soldered it together yourself and it included seven segment LEDs,
a keypad and a cassette interface running the Kansas City Standard.
You programmed in machine code in hexadecimal and I spent countless
hours creating programs and punching them in. What simultaneously got
programmed was my own brain as I can still cite many of the
6800 opcodes off-by-heart. 0x86 - Load A immediate,
0x7E - Jump extended, 0xCE load index register and so on. As useless
as still remembering your very first phone number.
Anyway, this computer had a profound impact and a little over a decade
later I found myself sitting in Silicon Valley in front of much more powerful
graphic workstations creating the test vectors for my first custom
integrated circuit.
The coolest computers I ever owned, which incidentally ran incredibly hot, were
a couple of Apollo DN1000 Personal Supercomputers.
See
http://jim.rees.org/apollo-archive/m...hure_Jul88.pdf
This same model computer was in use at CERN alongside their Crays.
64-bit RISC multi-processors with fabulous graphics. About 1000 were
made.
One of the oddest computers I ever worked with was the Intel iAPX 432.
It was Intel's first 32-bit architecture and were originally meant to
be the architecture that was to follow the 8008 and 8080.
The chipset included some of the largest integrated circuits ever designed
up to that time and only a very small number were ever made.
The instruction set was variable length and somewhat incredibly, bit
aligned.
Instructions could range from 6 bits to 361 bits long.
However, the programming language of choice for them was Ada.
Suffice to say they ran very slowly.
They were sufficiently complex to understand that the Intel
engineering documentation often resorted to using comic strips
to help succinctly convey some of the more subtle concepts.
Our primary research focus with them was fault tolerant computing.
A team at NASA who were also studying them for possible use on a then
future space station made up a significant part of a tiny user group.
), with tape drive for loading games, spent hours writing programs in basic, even had one used by the local special school to teach the kids arithmetic. Next one was a 286 "portable" 2 x 3 1/4" floppies, weighed about the same as a middy tower Pentium. 
