ARMAC was a one-off built at the Mathematisch Centrum with the main storage being a 56 track drum, but with a core memory with two-track capacity acting as a buffer, allowing it to operate as a bit-parallel machine, and perform 50 times faster than the predecessor bit-serial machine, ARRA II.
The ARMAC had two computing registers. The opcodes used 5 bits and addresses took 12 bits. Furthermore it was possible to use subroutine jumps, by using a link in a register. The computing machine consisted of 1200 tubes and consumed around 10 kilo Watt.
… a sub-routine was available to divide. Since division was very slow, most programmers decided to multiply with the inverse instead of performing a division. To illustrate the performance of the ARMAC:
Addition and subtraction of two numbers took 416 microseconds.
Multiplication of two numbers took 5.4 milliseconds.
Moving a track of 32 words took 14.6 milliseconds.
Regarding the inscription, “Wonder en is gheen wonder”, there are several suggestions for a translation, esp.,
the motto of early modern Dutch scientist Simon Stevin. It means something like: “It’s a miracle, but not a miracle”,
it’s more “imagine, then it’s not a miracle” or “think about it, then it’s not a miracle”. Miracle as in an unexplained event.
In arrangement and context, it also reminds of the inscription on Leibniz’s famous binary medal “Image of Creation” (for the Duke of Braunschweig-Wolfenbüttel, 1697), reading “Einer hat alles aus nichts gemacht / Eins ist noht.” (One has made all from nothing / [A] one is a nought.)
(To me, it would appear rather strange not to think of this medal as a reference.)
Expanding further on the similarity, the core memory diagram would go for the eye of God, the pentode for the Sun (it glows!) and the hysteresis diagram (of the magnetic cores) for the Moon (it has phases!). (At that time well the enabling triad of computing.) In this analogy, the two cross grid diagrams would serve as stand-ins for the binary core functions.
BTW, in early computers (e.g., the Univac I) instruction decoding was actually done by cross grid arrangements (i.e., function tables) and shift registers. This also applies for microcode sequencing. I don’t know anything about the ARMAC, but maybe… (If there’s any reference to Leibniz’s medal, at all, we may expect something with addition and multiplication.)
