Weekly Classic Computer Trivia Question (20141205)

Fred Cisin cisin at xenosoft.com
Sat Dec 6 15:17:38 CST 2014

> > NOTE:  In contrast with modern marketing, Seagate did NOT try to
> "maximize" the numbers by changing units nor using extraneous
> "significant" digits.
> They called the 506 a "5 Megabyte drive", NOT "5.33" nor "5.08".
> They called the 412 a "10 Megabyte drive", NOT "10.4" nor "10.16".

On Sat, 6 Dec 2014, tony duell wrote:
> I believe the last 2 digits of those model numbers was the unformatted
> capacity in some flavour of megabytes. As well as the ST412 (which had 2
> platters, 4 heads), there were the less common ST406 (one platter, 2
> heads) and ST419 (3 platters, 6 heads).  I guess it was a bit over 6
> megabyes per platter, so by the time you had 3 of them it rounded to 19
> megabytes unformatted. The ST406 and ST419 were used in the HP9133V and
> 9133XV respectively where the suffix letters helpfully give the
> formatted capacity in roman numerals.

Seagate was generally very good about providing the unformatted capacity,
as well as the formatted capacity ROUNDED DOWN.  They were even
cooperative about publishing the drive geometry.

Even unformatted capacity is potentially ambiguous.  In addition to the
basic geometry (number of heads and cylinders), it can further depend on
the spindle speed (function of the drive) the data transfer rate (function
of the controller), and encoding (function of the controller).
NEC, for instance ran their 3.5"HD disk at 360RPM, when almost everybody
else was using 300 RPM.

For floppies, there was encoding of FM, MFM, GCR, and a few others.
For a while, there were "RLL hard drives", which could be used as MFM with
good reliability, but corresponding lower capacity.  And there were
always a few power lusers who would attempt to use MFM drives as RLL, and
declare them "perfectly usable" if there were not immediate failures
(much like Spinrite restoring to use sectors that had been demonstrated to
be unreliable, if they passed one test).

Comparing drive capacities reliably required either knowing ALL of the
parameters, or making assumptions of some of them.

Grumpy Ol' Fred     		cisin at xenosoft.com

More information about the cctalk mailing list