R.S. wrote:
Bruce Hewson wrote:
Well Radoslaw,
As I said the original convention for disk was 1000 bytes = 1KByte.
No. There was no such convention. Possibly there was no convention at
all, although IMHO the most popular convention was IT k=1024, while
non-IT k=1000.
There was indeed a convention that pre-dated M$ and PC's by at least one
to two decades, established by IBM around the time of S/360 (1964) if
not before, and explicitly documented in their manuals: that for memory
K and M were to be used in the sense of today's Ki and Mi, while for
external storage (disk and tape) K and M were to have the standard Greek
prefix interpretations (extended to GB when it became affordable).
Since IBM was the major supplier of computer systems during that period,
IBM conventions, while not a standard, did carry much weight. It was
also well understood by most in the industry that the main reason for
making such a distinction was an architectural one, related to the use
of binary addressing for central memory, which dictated that physical
memory was installed in multiples of some appropriate power of two.
DASD and tape storage capacity had no such power-of-two constraints
because choice of physical DASD track capacity and tracks per device or
choice of physical tape block size and tape length could be arbitrarily
chosen.
IBM pretty well stuck with their convention until System Managed Storage
was introduced, and then suddenly DASD space was reported in "KB"
meaning KiB, and we got DASD allocation in "K" (=Ki) and "M" (=Mi)
records - but by that time the indiscriminate and capricious usage of
"KB" and "MB" by M$ and PC users was already in widespread practice.
Although M$ certainly institutionalized the confusion by reporting DASD
space in "KB" = KiB, I think M$ was to a large extent just reflecting
the imprecision and confusion of much of the PC User community who were
in general unaware of the decades of pre-PC computer history or the
architectural reasons why it might make sense for KB or MB to be
consistently interpreted differently in different contexts. The
Intel-based PC DASD restriction to 512 B sectors early in the game
contributed to this confusion by suggesting a hardware power-of-two
relationship for DASD architecture - although in this case dictated by
the Operating System convenience rather than by limitations of the media.
While there may not be any formal documented usage of "MB"= 1000x1024,
large base 10 numbers encourage our minds to division by powers of 10.
When confronted with an SMS free space value like 111,234 "KB" (=KiB),
it is instinctive to want to internalize this as ~111 KKiB, with the
ugly true units easily corrupted to "MB".
M$ did not comply with that convention when they decided to show disk
usage in binary format 1024bytes=1KByte.
M$ used k=1024 for disks for many years. Legacy MS-DOS (or rather IBM
PC-DOS!) used 1024 for many years.
Only memory (not disk) was specified in binary (1024) convention.
Again no. And again, you miss other (non-disk, non-memory) aspects like
tapes, networks, disk channels etc. etc.
And then other groups mixed 1000 and 1024 together when they reported
usage.
I have never met 1000 and 1024 mix, like 1*1000*1024=1MB. Neither in M$
nor elsewhere. Do you know such case ?
--
Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
