I've been mucking around with electronics kits from about Grade 5 (Electronics Australia was a staple. In 1974 Jim Rowe published the first build-it-yourself computer the EDUAC-8, but I was too young to buy or build it.) In 1976 I built a tennis video game project presented in Electronics Australia. The paddle joysticks varied the rate of charge on capacitors, that fed into schmitt triggers. This set the position of the bats on the screen. The capacitor was discharged on every screen refresh. The logic involved generating the ball movements, detecting collisions with walls/bat, and doing the appropriate reflections. All with 74C00 series logic. At Uni we mucked around with Sym's, TMS9900 (one of the first single-chip 16-bit CPU's), and so on.

About 1980; a vintage 6802 board. 2 x 6821 PIAs piggybacked to 4; 1K RAM biggy backed to 2K. On one of these we had 4 Ram chips piggybacked and an extra address docoder chip. Things progressed rapidly

This MC68705R3S contains just about everything the the board above has - on chip oscilator, 8-bit timer with 7 bit prescaler, 112bytes RAM, 3776 Bytes UV PROM, 4 channel AtoD converter, 24 IO lines. You had to erase it under a UV lamp for 10 minutes - thus the quartz window (soda glass won't pass hard UV) and the expense. I used one of these in the Rock saw. Now in 8 bit you can get (there's hundreds of combinations available) electrically erasable 60KB EEPROM, 2KB RAM, up to 24 channels of 10bit AD, SCI uart, SPI, I2C multiple 16 bit timers, PWM .... all in a chip. In the 2nd version pond controller I used a MC9S08QD4 - an 8 pin package that has 4KB Flash, 2KB RAM, 4 10bit AD's, 3 Timers, interrupts etc - in lots of 10,000 these cost $0.69 - and that was in 2007. I haven't built a circuit that doesn't have a MCU for years - even a simple sequencer to clean the 11 fountain jets, uses a MCU and 2 x ULN2803 darlington drivers.

In 1981 I bought a DEC LSI 11-23 with LP120 line printer and VT100 terminal. 64K Ram, 2 * 500KB 8" floppy drives, running RT-11. If a program was larger than 64K (any image maipulation was), then you had to do all the page swapping in your program, and try to arrange it at infrequently executed code so it didn't thrash the heads. Above are the chips from my original LSI11/23. Later I upgraded to another PDP 11-23, but with 512K ram and I think a 5MB hard drive. With the discovery of giant magnetoresistance by Grunberg and Fert in 1988, I now buy a 2,000,000MB hard drive for $100.00. And people have complained about funding the search for the Higgs!

Above is a 5Mb platter, In the middle is a 14bit * 14 word Core memory - yes the little black things are small toroids- some are missing.