Monday, April 28, 2014

The first practical RAM...

The first practical RAM... This is a short interview with one of the MIT engineers who helped develop magnetic core memories.  These were the first practical RAM devices, reasonably fast and reliable, unlike everything that came before them.

The first computers I ever worked with (Univac CP-642A) used exactly this kind of magnetic core bit plane memory – 30 planes, each with 32,768 bits.  They occupied about one cubic yard of the computer, packed with not only the bit planes themselves, but lots of both analog and digital circuitry.  They required constant adjustment (read and write current levels, pulse length and timing, and sense line gain), a tedious and involved procedure that was part art and part science.  In the U.S. Navy school that taught me how computers worked, and how to repair them, I also learned about older technologies: storage tubes and mercury delay line memories.  Weird stuff, by today's standards – but exotic and bleeding edge stuff back then...

Along the same lines, this video talks about “rope memory” the first practical ROM.  The first computers I worked on had something even more primitive: plugboard ROM, where the bits were programmed by placing pins in a “plugboard”.  Those computers had 32 words of 30-bit ROM, and somehow we got a bootstrap loader in that tiny bit of space!  Later I worked on computers that had capacitive ROMs, where the bits were programmed by the presence or absence of small circular areas on printed circuit boards.  Those ROMs had 64 words of 30-bit ROM – double the size.  Woo hoo!

The first microcomputer I built used a 1702 UV-erasable EPROM, with 256 bytes of ROM – more than the biggest mainframe ROM I'd worked on.  Those chips, as I recall, cost about $50 then – which seemed dirt cheap to me.  My first digital design project was a programmer for those 1702 EPROMs, so that I could enter two hexadecimal digits for each byte on a keypad and automatically program it.  Before I built this programmer, the only way I could “burn” code into a 1702 was to mail it off to a friend in San Diego who had a programmer, along with a hand-typed listing of the code I wanted.  When my ship was floating around in the Indian Ocean, the round trip could take several months!  I had a big incentive to build that programmer :)

No comments:

Post a Comment