Date:         Sat, 18 Apr 2015 09:32:38 -0400
Reply-To:     David Beierl <dbeierl@ATTGLOBAL.NET>
Sender:       Vanagon Mailing List <vanagon@gerry.vanagon.com>
From:         David Beierl <dbeierl@ATTGLOBAL.NET>
Subject:      Re: Friday..OT..Stuff from WalMart-like stores...
Comments: To: PB <pbrattan@GMAIL.COM>
In-Reply-To:  <CAOBs5F4Gi3SJ3Cn5CYncAORQHw8WemDKVV=jxEafOGn6DZq7iA@mail.g
              mail.com>
Content-Type: text/plain; charset="us-ascii"; format=flowed

At 07:11 AM 4/18/2015, PB wrote:
>Also, the actual usable capacity of a 128 GB hard drive is approximately
>only 112 to 114 GB!
>Patti

There are two different units called giga (mega, kilo, etc)
byte.  Semiconductor memory sizes are always powers of two because
it's built and addressed in a row-and-column fashion and each time
you exceed an even power of two in either actual storage or
addressing capacity, even by one byte, you have to double the
capacity; since it's there you might as well use it.  The same is
true of any value stored as bits.  One bit can store either of two
values, 0 or 1.  Two bits can store 00, 01, 10, 11.  Three bits 000,
001, 010, 011, 100, 101, 110, 111 and so forth.  Because powers of
two are so heavily wired into binary computing structure (exactly the
same way powers of ten are wired into decimal numbering systems),
it's convenient to think of "kilo" as 2^10 rather than 10^3; or 1024
vs 1000.  LIkewise "mega" represents 1048576 instead of 1000000 and so forth.

But disk space is essentially a linear quantity and there's no
physical reason you couldn't add capacity in decimal numbers of bits;
so historically drive mfrs have quoted capacity using 10^3, 10^6 etc.
instead of 2^10, 2^20 and so forth.  And that is why a megabyte of
memory won't fit into a megabyte of disk.

Actually in the early days of "Winchester" type fixed hard drives
(named after an IBM behemoth that had 30 megabytes fixed and 30
megabytes removable storage), drive mfrs like Shugart, Tandon,
Micropolis etc quoted unformatted capacity, the actual number of raw
bit locations available on the disk surface.  Since there's call it
30% overhead involved in formatting the data into locatable chunks
called sectors, that is a highly misleading number.  They could (just
barely) get away with it because in theory you could devise any
formatting scheme you liked.  In practice it was a huge pain and I'm
sure everyone was relieved when formatted capacity became the quoted
number.  Nowadays the formatting scheme is built into the drive
itself and the drive systematically lies to the outside world about
how the data is actually physically stored.  To the outside world
it's presented as so many tracks each containing so many seof 512
bytes each, and what the mfr does internally is somewhat beside the
point.  Given the storage densities available today it is very likely
black magic.  I wouldn't stand too close to magnetic storage
engineers during lightning storms.

Yours,
David