https://nosher.net/archives/computers/your_computer_1987-11_001

Catt amongst the pigeons

Sinclair's wafer-scale memory devices were developed from the 1978 designs of Ivor Catt, a man who had made some enemies over the years by suggesting that the classical Von Neumann architecture of computers - where a central CPU controlled all the memory, did all the calculations and managed input/output, was essentially "bunk". After several years maintaining his ideas at places like Middlesex Polytechnic and running digital electronics courses, he popped up again at Sinclair. Sinclair's product - essentially a solid-state Winchester disk - wasn't quite the culmination of Catt's dream, but at least it was using his technology, albeit using blocks of RAM instead of the original serial memory design that Catt had in mind.

The fundamental problem of producing a single silicon wafer was that imperfections would mean that certain parts of the chip wouldn't work, which would normally mean rejecting the entire wafer. This was solved for conventional chips, which were cut out in their tens from a single wafer of silicon, by simply discarding any chips that didn't work, however Catt's Content Addressable Memory solution allowed the wafer to determine for itself which were working sections of memory by injecting test patterns into each section and seeing which could be read again, meaning not only that in more cases the entire wafer was usable but that the entire thing was inherently fault tolerant^[30].

The first units from Anamartic, expected at the end of 1988, were to be CMOS, but Sinclair saw the development of bipolar technology as crucial, as this required about one twentieth the voltage to drive the bit-switching transistor. The whole idea was really 25 years ahead of its time, with the rise of solid-state disks based on various forms of Flash memory only becoming commercially viable from around 2005.