Microcomputers As A Space Spinoff 3842

On Mon, 3 Jul 2006 20:09:30 +0000 (UTC)

The first mainframe semiconductor memories were built with 1103s and in 1970 I think that was pretty much the only market for dynamic RAM.

Core certainly wasn't displaced overnight, it had advantages over semiconductor memory (non volatile springs to mind) and may well have been more reliable than early semiconductor memories, but the 1103 was used in places where previously core was the only option.

I'm not sure how much the microcomputer had to do with the displacement of core, it would be about 1977 that I saw my first semiconductor megabyte connected to a 370.

Mainframes started to use semiconductor memory starting with the 1103, no micro ever used the 1103.

It was already dying by then.

Mainframe memory banks started to be made with dynamic RAM instead of core and there came a time when core was no longer used for mainframes, that was the sense I used the word in.

Indeed - but note that SRAM wasn't around for long before DRAM and it was DRAM that sold as a core replacement not SRAM. Indeed it was probably the microprocessor that rescued SRAM from an early rest similar to that experienced later by bubble memory.

The Z80 ran at 2.5MHz it was the Z80A that pushed it to 4MHz (and the Z80B pushed it to 6MHz). The refresh time overhead was never onerous but the logic (counter, address multiplexer and logic to hold the CPU off the bus ate board space). The Z80 included the logic for managing the refresh cycle for DRAM which removed the need to provide a refresh controller. Reducing chip count was a key factor in reducing costs in those days.

IIRC the last two were pretty close to simultaneous.

-- The computer obeys and wins. A better way to focus the sun You lose and Bill collects. licences available see