Please see comments below:
Dick
-----Original Message-----
From: allisonp_at_world.std.com <allisonp_at_world.std.com>
To: Discussion re-collecting of classic computers
<classiccmp_at_u.washington.edu>
Date: Thursday, May 27, 1999 7:02 AM
Subject: Re: non-SCSI disks on a SCSI disk interface (was Re: Space, the
next frontier)
>> > The 2190 is 1224*15. However the number of cylinders can be increased
(it
>> > was controller limit not the drive) the limit, I think is closer to
>> > 4096*16*18*512 or roughly over 600mb. The problem is that RLL with the
>>
>> Do you mean that Maxtor used only about the first third of the available
>> platter for data? That seems like a bit of a waste. I do recall - years
>> back - discussion about how the XT1085 could be extended, but not by a
>> factor of three!
>No, No No...
>
>The maxtor 2190 is 1224 Cylinders by 15 heads and whatever formatting for
>the number of sectors.
This lies within the capacity limit of 504 MB imposed by many older '486
motherboards.
>the command structure for most MFM and RLL controllers only used
>1024 of the possible 4096 (10 of the available 12 bits). It's an artifact
>of the controller designs.
Where do these 12 bits fit? The 10-bit limit was inherited from the WD1010
chip's register geometry.
This was frequently a motherboard/BIOS-imposed restriction. If the
Motherboard doesn't recognize drive types with more than 1024 cylinders, the
controller won't either. Several controllers offered various ways around
this, e.g. masquerading fewer heads up to the maximum of 16 and reducing the
number of cylinders correspondingly, and, likewise, fiddling with the number
of sectors so that it was up to the BIOS-compatible maximum of 63 whereas
the drive only had 15 heads, again with a corresponding reduction in the
number of cylinders. However, if the BIOS on the PC motherboard supports
drives up to but not exceeding the 1024x16x63x512-byte configuration, then
that's all it will do.
>Physically you can go for up to 16 heads and 4096 cylinders in the st506
>interface spec'd drives. The problem was that more cylinders do not
>enhance speed. It's better to pack more on a given cylinder as you can
>access that data faster than moving heads across a lot of tracks.
Please explain what you mean by this. I've not yet gotten feedback from my
ex-girlfriend's son, currently chief eng'r at Maxtor, but I've seen no
information which suggests that the 2190 drive actually had 4K cylinders.
You must be referring to some geometry aliasing stunt performed by some
controllers. I'm mystified. I can see no way a controller can circumvent
the BIOS, which is the vehicle by means of which the OS communicates with
the HDD at the lowest level. The only exception I've run into is the Lark
Associates controller, which lies to the motherboard about what it attached
and presents, say, my 8760 as two physical drives, of which one is at the
maximal 504 MB capacity, while the remainder of the drive appears to be on a
separate physical drive.
>Please everyone prune of the unwated portions of the message. If I can do
>it using pine across a sluggish telnet link during the day No one has an
>excuse.
>
>Allison
>
>
Received on Thu May 27 1999 - 09:29:35 BST