32-bit floats, 7 bit exponent, 24 bit mantissa, with an extra bit for signifying "special values" (infinity/NaN/0/~0). Fixed-length 8 bit instructions. Two registers. Separate load/store, arithmetic is reg-reg only. Branch delay slots (5 § 1).
RISC forty years before RISC. Zuse was and is an utterly underappreciated genius.
Fascinating. From page 8, §11, in the section on "Unterpläne", it looks also as if it used a kind of function calls and subroutines. "Plan" seems to mean "program", so "Unterplan" is literally subprogram, and it had a facility to automatically return to the point where it the subprogram was called.
>While working on his doctoral dissertation, Zuse developed the first known formal system of algorithm notation[7] capable of handling branches and loops.[8][9]
The Z4 was the successor of the Z3 which was the world's first working programmable, fully automatic digital computer. Both machines were invented by the German engineer Konrad Zuse from 1935-1941 (Z3) and 1942-1945 (Z4).
During the latter period, Zuse also created "Plankalkül", the first high-level programming language to be designed for a computer.
Clear text. Simple, coherent typesetting with underlined headings. Once so common: now, rarity of rarities. You can see the connection with the layout of man pages through groff...
It may not be obvious to HN'ers who grew up this century, but while some of the images (such as the illustration of film holes for the opcodes, K1.§4) and underlining was done on the typewriter, many of the graphics and special symbols (e.g. K1 hero image, or I/O, K2.§6) would then be added by hand in a second pass, à la MathJax.
Am I reading K4.§1 correctly that intermediate values were kept to about 80 bits?
I am fond of the hack of K3.§5.4. We unroll loops, they rolled them.
Especially important: K5.§1: bugs and workarounds!
I think all of them are important but clearly Konrad Zuse has received way less attention than he deserves. Growing up I remember being taught that ENIAC was the first computer and didn’t even hear about the Z1 to Z4.
But reading up on the details I find that utterly unfair. ENIAC e as way too far away from a programmable machine. It was computation units which just got wired together in different fashion. That is very removed from what a computer is today. The Z1 and later is far closer to what we think of as a computer IMHO, with actual programs being fed into it decoded using registers and and ALU etc.
RISC forty years before RISC. Zuse was and is an utterly underappreciated genius.