home *** CD-ROM | disk | FTP | other *** search
- Path: fc.hp.com!tomk
- From: tomk@fc.hp.com (Tom Kennedy)
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: Zip Drive Geometry
- Date: 28 Feb 1996 22:08:30 GMT
- Organization: Hewlett-Packard Fort Collins Site
- Message-ID: <4h2jou$fug@fcnews.fc.hp.com>
- References: <4glcie$b66@fcnews.fc.hp.com> <4gs5ab$alg@eken.hv.se> <1824.6631T598T423@wvlink.mpl.com>
- NNTP-Posting-Host: magnum.fc.hp.com
- X-Newsreader: TIN [version 1.2 PL1.4]
-
- Gene Heskett (gene_heskett@wvlink.mpl.com) wrote:
-
- : HW> In article <4glcie$b66@fcnews.fc.hp.com>, tomk@fc.hp.com (Tom
- : HW> Kennedy) writes:
- : >>David Evans (devans@ccubb.com) wrote:
- : >>: Does anyone know the geometry of a ZIP disc? I seem to have the
- : >>: wrong layout and format doesn't work...
- : >>
- : >>It's got 196608 sectors. You can pick any geometry that has that
- : >>number of sectors. (I use 8 sec/track, 2 heads, 12288 tracks.)
-
- : HW> Why so few secors and so many tracks? Isn't that a major
- : HW> performance loss? The number of seeks for a track increases a
- : HW> lot, and that time is a lot higher than the time required to
- : HW> revolve the disk one rotation. Just my humble opinion...
-
- : HW> /Henrik
-
- : Henrik;
-
- : It normally doesn't make any difference to a scsi device since the
- : internal physical disk format is not something we have had any
- : control over in quite some time now.
-
- : The track and sector numbers are normally chosen so that they
- : duplicate as close as possible, without going over, the actual
- : capacity of the disk. The controller in the computer asks for a
- : sector based on tracks*sectors_per_track*head, commonly called the
- : Logical Sector Number, and the controller on the drive then
- : translates that often quite large number back into the actual
- : location on its platters. We, the users, have no way of accessing
- : what may be on the drives actual track 117, sector 3, head 5 except
- : by pure dumb luck! Thats what he meant above when he said to pick
- : any set of numbers that totalled up to 196608 sectors. The actual
- : Logical Sector Number (LSN) is the only address that in fact goes up
- : the cable to the drive.
-
- Exactly.
-
- And I did some tests with DiskSpeed with different geometries (in case
- there was some difference with the Amiga filesystem -- I didn't think
- there would be, but it was easy to test...). As I expected, there was
- no preformance difference with different # sec/track.
-
- The reason I choose 8 sec/track specifically was to optimize the
- number of sectors available to the filesystem. One cylinder must be
- used for a RDB. Since the RDB only needs 2-3 sectors for a Zip disk,
- I picked the smallest sec/track possible that still came up with a
- geometry that has 196608 sectors even. (You could do less then 8
- sec/track, but then the number of cylinders is too big for HDToolBox.
- The current version uses 16-bit ints for cylinders on the partitioning
- screen. It should use 32-bit ints...)
-
- My brother and I calculated all this stuff precisely to minimize the
- number of unuesd sectors on a disk. 8 sec/track is the best you can
- do :)
-
- (Ok, some people might not care about wasting a few K here and
- there... We figured it wasn't that much work, and adds up across
- dozens of disks.)
-
- Tom Kennedy
-