1311 vs 2311 disk capacity 2475

Sarr J. Blumson

Well, I wouldn't trust my memory either. :-) But I believe the 1620 only had 4 data bits, or at least only 4 that were visible. It could only represent 0-9; characters required 2 digits.

Each digit of memory was six bits:

the check bit (what would normally be called a parity bit), the flag bit,

An 2-8 (the "two" and "eight" bits) signififed a group mark and was used to terminate I-O transfer. The group mark may or may not have the flag bit on.

And you are correct, to represent a character, two decimal digits were used. There were instructions to modify the flag bit and to create a group mark, but other than that, the individual bits weren't adressabile or setable, other than setting the complete "digit". Storage was one of: 20k, 40k, or 60k (for a model I). The model I used specific low storage locations to hold a table for addition and multiplication. The model II removed that necessity of a lookup table.

Most instructions were 12 decimal digits, with a few 7 decimal digit instructions. While as a student, my univerisity had a full FORTRAN II (we called it 2.5) that ran in about 18k decimal digits, but required 40k to actually compile a decent program. It did the compile in one pbutt as the IBM 1620 had indirect addressing. For the most part, the compiler ran at reader speed included driving the printer and punching an (excutable) object deck. Gerard S.