The ANFSQ31 Did Exist

(John Savard) wrote, in part:

Of course, I was thinking of the famous post by Robert Firth from comp.arch:

**

Indeed it was. Here is one list, from the KDF9 programming manual, p 24:

THE KDF9 WORD HAS 48 BITS ...

IT MAY BE USED AS...

Eight 6-Bit Alpha-Numeric Characters One 48-Bit Fixed-Point Number Two 24-Bit (Half length) Fixed-Point Numbers Half of a 96-Bit (Double Length) Fixed-Point Number One 48-Bit Floating-Point Number Two 24-Bit (Half length) Floating-Point Numbers Half of a 96-Bit (Double length) Floating-Point Number Three 16-Bit (Fixed point) Integers Six 8-Bit Instruction Syllables

An instruction was 1, 2 or 3 syllables; an address was 15 bits. O, memory! We shall not see its like again.

**

The advertisement from the KDF9 brochure (p. 10-11) looks like this:

**

-- Eight 6-bit alpha-numeric characters One 48-bit fixed-point ------------------ number T H E Two half-length fixed-point numbers K D F 9 IT Half double-length W O R D MAY fixed-point number H A S USED One 48-bit floating-point AS number 4 8 Two half-length B I T S floating-point numbers ------------------ Half double-length floating-point number --

**

Of course, the KDF 9 was a computer to inspire pbuttionate devotion.

Thus, from

I ran across (presented here with slight edits)...

**

Some talk of I.B.M.s and some of C.D.C.s, Of Honeywells and Burroughses, and such great names as these, But of all the world's computers, there's none that is so fine, As the English Electric Leo Marconi Kay - Dee - Eff Nine!

Some talk of thirty-two bit words, and some of twenty-four, Of disks and drums and datacells, and megabytes of core, But for those who've written usercode there's nothing can outshine, The subroutine jump nesting store of the Kay - Dee - Eff Nine!

**

I see I may have been overly harsh in denying it a capability for 16-bit arithmetic; I had remembered seeing something about a nesting store containing 16-bit addresses, but I thought that this was just for subroutine returns.

While there was one general nesting store for 48-bit values, and another nesting store for subroutine return addresses, it was the Q-store that had sixteen values, each 48-bit value divided into three sixteen-bit parts, counter, increment, and modifier. And these three parts could be manipulated independently. Addresses occurring within programs, those being composed of eight-bit instruction syllables, were also 16 bits in length.

Given that the machine provided capacities for address arithmetic as well as indexing, it certainly did handle 16-bit integer values. Being from the beginning of the 1960s, before the IBM 360 splashed into the world of computing, it would be unfair to criticize it for not supporting 8-bit characters.

But it is those few computers that did support both the 8-bit and the 6-bit character that made the fullest use of the potential of the 24-bit or 48-bit word. The KDF 9 has many *other* merits, and if it can't really fully lay claim to this particuar one, although touching on it to a greater extent than I thought, that does not greatly diminish its importance.

And, of course, when it comes to variable character lengths, the STRETCH, with a 64-bit word, used bit field instructions for all its fixed-point operations (so that people in a hurry to multiply integers used unnormalized floating-point instead), or the PDP-10, which let one divide its 36-bit word into bytes of any size, must take top honors.

Even though, unlike the Honeywell 200, these computers had word lengths which *did not* in themselves suggest flexibility; a 64-bit word is a pure power of two, and a 36-bit word is generally broken down into 18, 12, and 6 bit pieces; nobody but the PDP-10 thought of 9-bit pieces, let alone putting five 7-bit characters in a word with only one bit left over.

John Savard Usenet Zone Free Binaries Usenet Server More than 140,000 groups Unlimited download