home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.hardware
- Path: sparky!uunet!spool.mu.edu!agate!linus!linus.mitre.org!jcmorris
- From: jcmorris@mwunix.mitre.org (Joe Morris)
- Subject: Re: The maxtor 213 meg drive is NOT 213 megs!
- Message-ID: <jcmorris.725219398@mwunix>
- Sender: news@linus.mitre.org (News Service)
- Nntp-Posting-Host: mwunix.mitre.org
- Organization: The MITRE Corporation
- References: <s106275.725140761@ee.tut.fi> <1992Dec24.103209.21401@pasteur.Berkeley.EDU> <BzrLA1.D4C@cs.vu.nl> <1992Dec24.163552.3703@cc.gatech.edu>
- Date: Thu, 24 Dec 1992 17:49:58 GMT
- Lines: 57
-
- duggan@cc.gatech.edu (Rick Duggan) writes:
-
- >And all this talk of UNformatted capacity is driving me nuts. Exactly
- >how much data can you store on an UNformatted drive? Sound like
- >0 bytes to me :-). If you gotta format it to use it...
-
- If you're limiting your discussion to PC-compatible environments, the
- unformatted capacity of a drive is not particularly meaningful except
- as a first-order approximation, to be used in comparison to other drives'
- unformatted capacity spec.
-
- If you are referring to generic environments, or to computer systems which
- allow the user to specify the physical record lengths of data on a disk,
- then the "unformatted" capacity is a valid measure of the *maximum*
- amount of data which can be stored. Until the length of the recorded
- data blocks is specified, it's the only valid measure of the disk capacity.
-
- Of course, you will achieve the maximum value only if every track on
- the drive is formatted to contain exactly one record, and that record
- has a length equal to the entire track. Whenever you start formatting
- a track to contain more than one physical record, you will lose some
- of the capacity of the drive. The track will still have the same number
- of bytes of data, but some of them are now allocated to the inter-record
- gap (IRG), others to the record headers used by the disk to locate the
- data records, and still others to checksums and other administrative
- overhead. The number of bytes lost in this way is a function of the
- design of the disk and its controller.
-
- As an example, most IBM mainframe disks allow the user to specify the
- physical blocksize of the data on the disk. For a 3380 disk, each
- track has a maximum unkeyed record length of 47476 bytes...but look
- at what happens when you start subdividing the real estate:
-
- Record # of equal-size Usable
- size records per track bytes
- -------- ----- ------
- 47476 1 47476
- 9076 5 45480
- 4276 10 42760
- 2164 18 38952
-
- Formatting the track for 512-byte records would allow 46 records per
- track, for a track capacity of 23552 bytes...about a 50% loss from the
- unformatted capacity.
-
- The comparison isn't completely valid because the mainframe disk has
- to provide functions not necessary on a desktop system where the
- block (= "sector") sizes are immutably fixed, but you can see that
- the overhead associated with each block can quickly consume a lot of
- the "unformatted" capacity when you format for small record sizes.
-
- Now, if we can convince MS to allow DOS to talk (in native mode) to a
- physical disk whose sectors are equal to the cluster size (say, 2048
- bytes for a typical hard disk today) then the amount of usable space would
- go up by some amount for a disk of given unformatted capacity.
-
- Joe Morris
-