Byte

[Peter Flass]

The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf

The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.

[Peter Flass]

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.

Wikipedia has a decent article:
https://en.wikipedia.org/wiki/Byte

I’ve seen “syllable” used on the B-5500.

[Scott Lurndal]

Post by Peter Flass

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.

https://en.wikipedia.org/wiki/Byte
I’ve seen “syllable” used on the B-5500.

And on the B3500, both of which originated in the old Electrodata
plant in Pasadena where the B300 preceeded both. For the B3500,
syllables referred to the instruction operands (i.e. A syllable,
B syllable or C syllable).

[Dan Espen]

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.

Nice work finding that and the Wikipedia article.
Wikipedia puts the date as 1956.

[Timothy McCaffrey]

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.
--
Pete

IIRC, CDC with the 6000 series used, character to mean 6 bits, byte was 12 bits, and a word was 60 bits.
The PPs addressed memory as "bytes" (12 bits).

Yes, the Unisys MCP systems still use syllable for 8 bits (because instructions are 8 bit granular),
they started off as 6 or 8 bit character systems (6 for "field data" IIRC, 8 for EBCDIC, 7 bit ASCII was
added later, but used 8 bits). A word is 48 bits. Started with a 3 bit tag, later 4 bits and now 16 bits.

- Tim

[Peter Flass]

Post by Timothy McCaffrey

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.
--
Pete

IIRC, CDC with the 6000 series used, character to mean 6 bits, byte was
12 bits, and a word was 60 bits.
The PPs addressed memory as "bytes" (12 bits).

That would be central memory, I assume. Perhaps because PP word size was (I
believe) 12 bits.

Post by Timothy McCaffrey
Yes, the Unisys MCP systems still use syllable for 8 bits (because
instructions are 8 bit granular),
they started off as 6 or 8 bit character systems (6 for "field data"
IIRC, 8 for EBCDIC, 7 bit ASCII was
added later, but used 8 bits). A word is 48 bits. Started with a 3 bit
tag, later 4 bits and now 16 bits.
- Tim

[Mike]

On Nov 11, 2022, Peter Flass wrote
(in

Post by Peter Flass

Post by Timothy McCaffrey

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.
--
Pete

IIRC, CDC with the 6000 series used, character to mean 6 bits, byte was
12 bits, and a word was 60 bits.
The PPs addressed memory as "bytes" (12 bits).

That would be central memory, I assume. Perhaps because PP word size was (I
believe) 12 bits.

Post by Timothy McCaffrey
Yes, the Unisys MCP systems still use syllable for 8 bits (because
instructions are 8 bit granular),
they started off as 6 or 8 bit character systems (6 for "field data"
IIRC, 8 for EBCDIC, 7 bit ASCII was
added later, but used 8 bits). A word is 48 bits. Started with a 3 bit
tag, later 4 bits and now 16 bits.
- Tim

If memory serves me correctly, the OO’s er PDP-8’s 12 bit machines.

Michael LeVine
***@redshift.com

Politics is the art of looking for trouble,
finding it everywhere,
diagnosing it incorrectly,
and applying the wrong remedies.
Groucho Marx

[Fred Weigel]

Post by Peter Flass

Post by Timothy McCaffrey

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.
--
Pete

IIRC, CDC with the 6000 series used, character to mean 6 bits, byte was
12 bits, and a word was 60 bits.
The PPs addressed memory as "bytes" (12 bits).

That would be central memory, I assume. Perhaps because PP word size was (I
believe) 12 bits.

Post by Timothy McCaffrey
Yes, the Unisys MCP systems still use syllable for 8 bits (because
instructions are 8 bit granular),
they started off as 6 or 8 bit character systems (6 for "field data"
IIRC, 8 for EBCDIC, 7 bit ASCII was
added later, but used 8 bits). A word is 48 bits. Started with a 3 bit
tag, later 4 bits and now 16 bits.
- Tim

--
Pete

12 bits b/c a card has 12 punch rows

[Anne & Lynn Wheeler]

Post by Fred Weigel
12 bits b/c a card has 12 punch rows

pre-360, bcd was alphanumerica, small subset of 12 punch row
combination.

compatibility with pre-360 cards was "column binary" ... two six-bit
"bytes" per column. read/write on 360 ... would (column binary) i/o card
with 160 length ... two 360 bytes per column (using just 6bits/byte).

univ. had 709 tape->tape with 1401 "unit record" front-end running MPIO
(tape->printer/punch, card read->tape). They were sold a 360/67 for
TSS/360, replacing 709/1401. Temporarily pending arrival of 360/67, the
1401 was replaced with 360/30 (which had 1401 emulation mode). I had
just taken two credit hr intro to fortran/computers and at the end of
the semester I was hired to rewrite 1401 MPIO in 360 assembler (running
360/30 as 360 instead of as 1401). The univ. shutdown the datacenter on
weekends and they gave me bunch of software & hardware manuals and I
would have the datacenter all to myself over the weekends (although
48hrs w/o sleep made monday classes hard). I got to design & implement
my own monitor, device drivers, interrupt handlers, error recovery,
storage management, etc. Within a few weeks had 2000 card 360 assembler
program.

Later, I assumed it was just exercise in somebody learning 360 (since
1401 MPIO ran just fine on 360/30 in 1401 emulation). Within year of
taking intro class, 360/67 came in (replacing 709 & 360/30), TSS/360
never came into production, and I was hired fulltime responsible for
OS/360 (and continued to have my dedicated weekend 48hr datacenter
time).

2540 reader/punch
https://en.wikipedia.org/wiki/IBM_2540

2540 trivia: univ. class registration was card based. 2540 had five
output stackers, two dedicated for reader, two dedicated for punch and
middle stacker could feed from both reader and punch. program processing
cards would read into middle stacker. punch feed was was loaded with
colored striped cards ... if a problem was found with registration card,
a blank card would be "punched" into middle stacker behind problem
card. There was more than a dozen card trays (2000+ cards/tray). After run
would pull each registration card that had a blank colored card behind
it (for further processing).

[Louis Krupp]

<snip>
Yes, the Unisys MCP systems still use syllable for 8 bits (because instructions are 8 bit granular),
they started off as 6 or 8 bit character systems (6 for "field data" IIRC, 8 for EBCDIC, 7 bit ASCII was
added later, but used 8 bits). A word is 48 bits. Started with a 3 bit tag, later 4 bits and now 16 bits.

I remember "syllable" being used to describe instructions on the B5500
(if I recall correctly, each instruction used 12 bits, while each
character used 6, and each word contained 48 bits), but not on the B6700
or its successors, where instructions used (and presumably still use) a
variable number of bytes.

The B6700, etc had a translate table to convert EBCDIC to ASCII and back.

I didn't know there were now 16 tag bits. The B5500 had no tag bits, but
bit zero was a flag bit, and the only way to manipulate something with a
flag bit set was with stream procedures. I was about 16 when I learned
about this stuff and I thought stream procedures were just the coolest
thing.

Louis

[Vir Campestris]

Post by Peter Flass
The word may slightly precede S/360. Here is one article from 1960 talking
about the “Harvest” system that uses it (p.23)
http://www.bitsavers.org/pdf/afips/1960-05_%2317.pdf
The Google ngram viewer is useful for determining the date of appearance of
a word, or would be if the publication dates weren’t often wrong. Between
that and being unable to view the title page of many books and papers to
verify the date, one reference is all I had the patience for, but there
appear to be others, but not much from before 1960.

Over in the other thread Bob pointed me at

http://pdp10.nocrew.org/docs/instruction-set/Byte.html

where on the PDP-10 a byte was an arbitrary number of bits.

Andy

[John Levine]

Post by Vir Campestris
where on the PDP-10 a byte was an arbitrary number of bits.

Right. The original byte machine was the IBM STRETCH, which could
address bytes of arbitrary size. The slightly later 4020, used in the
SAGE air defense system, had 48 bit words and 6 bit bytes. S/360 is
as far as I know the first machine with 8 bit bytes and the first
with byte addressing, with larger data units at 2, 4, or 8 byte
intervals.

The PDP-6 and -10 were word addressed but had load and store byte
instructions that could load and store bytes at any size and offset
that would fit in a 36 bit word, and could step through a byte string,
e.g., ILDB advanced the pointer to the next byte and fetched (Loaded)
that byte. Late versions of the PDP-10 had microcoded string
instructions that used the same byte pointer format.

[Quadibloc]

Post by Peter Flass
The word may slightly precede S/360.

Oh, indeed. IBM coined that word, for the STRETCH, of which
Harvest was a variant. On that machine, a byte was a variable
number of bits in length, and could cross word boundaries.

On the PDP-10, a byte was also a variable number of bits in
length, although it could not cross word boundaries, making it
the closest spiritual successor of the STRETCH in this
respect.

But most of the world learned of the word from the System/360,
and so its definition became the most popular one.

John Savard