[MEncoder-users] Using SI units in mencoder

[asym]

At 09:27 2/9/2005, Trent Piepho wrote:
>On Wed, 9 Feb 2005, asym wrote:
> > At 06:40 2/9/2005, Trent Piepho wrote:
> > >On Wed, 9 Feb 2005, asym wrote:
> > > > The is just so much BS.  The use of kilobyte to mean 10^3 bytes was 
> around
> > > > long long long before software developers and self righteous 
> technophiles
> > >
> > >Really?  The datasheet for the 4004 microprocessor,
> > >http://www.piercefuller.com/collect/i4004/, from 1971 uses K to mean 
> 1024 when
> > >referring to numbers of bytes or bits.  Do you have a reference to 
> kilobytes
> > >meaning 1000 bytes that predates that?
> >
> > Try every storage device or mainframe from IBM produced in the roughly 15
> > years prior to the 4004.  All of these, like the RAMAC (~1956), Ramkit, 
> etc
> > predate the microprocessor, not just the 4004.  All of them listed
>
>Really?  Here is the manual for the IBM 32K Assembler for the System/360, from
>1966:
>
>http://www.bitsavers.org/pdf/ibm/360/Y26-3598-0_32k_asmPLM_1966.pdf
>
>   "The Assember operates ... on System/360 Models ... with at least
>   32,768 bytes of main storage."
>
>Here 32K quite clearly means 32 * 2^10 bytes.
>
>While I've seen the RAMAC described as having a storage capacity of 5 million
>7 bit characters, I've not seen anything from that era referring to this as "5
>megabytes".  Can you find actual text from the 50s where kilobytes are 1000
>bytes?

I'll look, I don't have anything nearly this old in my personal posession, 
but I'll see if some intrepid soul has scanned or transcribed this stuff 
and put it on the web somewhere I can find.

IBM and some other companies had many different mass storage devices that 
predated (for computers at least) the transistor, and most of these were 
not power-of-two-bound.  Magnetic film/tape, core memory, vacuum tubes, and 
CRTs were all used for data storage before the magnetic disk, and obviously 
well before the microprocessor or the IC.

I could have to eat my words, because I'm not sure when the term "byte" 
came into the lexicon, and prior to byte they used "word" and "character" 
to define storage capacity.

Regardless, both Tetsuos argument and mine both stand.  His is simple -- 
mplayer/mencoder don't document what definition they are using for 
kb/kB/KB/etc.

Mine (and others) is that unless they are using the SI definition, they are 
in fact "wrong."  Kilo is a prefix with a specific meaning, a meaning it 
has had for thousands of years, a meaning accepted and standardized in all 
professional fields throughout the world with a singular exception -- the 
microprocessor addressing world.  Taking the prefix kilo (thousand), 
arbitrarily redefining it to mean 1024, and using it IS the same thing as 
redefining "gallon" to mean 5 quarts instead of 4, just because it suits you.

The IEEE did NOT simply cowtow to "hard drive manufacturer marketing 
departments" as you stated.  The meaning "kilobyte = 2^10" has been around 
longer than the IEEE terminology as you said, however, the meaning "kilo = 
1000" has been around longer still, and is far more widely accepted and 
understood than either.

There is one codified standard by an international standards body that 
defines what "1024 bytes" is.  That standard is IEC 60027-2 (IEC not IEEE, 
my mistake) and the nomenclature is "kibibyte".  Why do so many people 
resist this change, when all it requires is getting into an easy new habit, 
and will remove all the ambiguity that started this whole 
discussion?  Because they're stuck in their ways, resistant to change, 
lazy, elitist, full of conspiracy theories -- pick your cliche.

For my part, I also advocate using the term "octets" instead of "bytes" -- 
octet isn't ambiguous, byte and word both are, but that is a discussion for 
another day, though machines (microprocessor architectures anyway) that do 
not use 8-bit bytes are still in use today.