PC Parity, was identifying PC Simms

[Chuck Guzis]

Parity actually made a bit of sense before the adoption of the SIMM/SIP memory module.  There were a couple of diagnostic tools that would draw a graphical representation of the memory array and point to the failing chip.   Parity detection, at least in theory, allowed quick detection of the failing bank before much data became corrupted.  

But it was far from a complete solution--there were too many weak links in the chain.  Hard disk controllers, for example never used parity checking on their on-board buffers.  And the bus to the adapter cards wasn't parity checked.

Back in the mid 80's, a friend and I were tasked with determining if the commodity PC was suitable for critical control applications.  There were too many points of vulnerability in the PC design, so we worked out a majority-polled 3 PC setup connected by a high-speed redundant serial link.  We had hot-swapping and dynamic resyncing capabilities.  As a concept model it was pretty impressive to watch operate, but never really went anywhere.   And, as you might imagine, performance wasn't wonderful.

As someone's already mentioned, however, it's too bad that the socketed positions on the PC used low-quality sockets that really couldn't take too many insertions before developing problems of their own.

I seem to recall some do-it-yourself SIMMs made in the 80's by Jameco or JDR that provided the "glue" on a socketed PC board.  You could plug in your own DIP memory chips to make a SIMM.

Cheers,
Chuck