[net.arch] IBM 1130 mini had regs in memory too

gnu@sun.uucp (John Gilmore) (12/10/83)

It had a "real" accumulator and multiplier/quotient register, but
the "index registers" were in core locations 1, 2, and 3 (this was in
the days of real "core" core, too, with 3.6us cycle times).  Its sister
machine, the 1800 process-control computer, had the registers in real
registers, so if you wanted your code to be portable you couldn't
take advantage.  Not that there was any advantage anyway...