28 Oct
2000
28 Oct
'00
7:56 p.m.
"Richard Erlacher" <richard(a)idcomm.com> wrote:
I don't see how switching from a 1010 to a 2010
will help with the number of
heads in a controller => drive interface. Both chips have three bits to
define the head address. One thing that could be done with the external SDH
register, however, is encode a drive select bit as a head address.
Correct. The only reason to use a WD2010 in an AT&T 7300 or 3B1 is that
the WD2010 supports up to 2048 cylinders, while the WD1010 only supported
1024. If you use a Maxtor XT2190 or equivalent, which as 1224 cylinders,
you lose 200 cylinders if you only have the WD1010.
The hack to support more than eight heads (such as 15 for the aforementioned
XT2190) did use external logic to produce the most significant head select
line. This had the consequence that the address headers on the disk for
heads 0 and 8 were identical, thus if you tried to read with head 0 but
the most significant head select line was stuck high, it would happily
read from head 8 instead with no way to tell that it was getting the
wrong data. Similarly for the other six cases, and for writing.
IIRC, Intel second-sourced the WD1010 and WD2010 as the i82062 and i82064.