On 20 Dec 98 at 19:48, William Patton wrote:

> I am trying to decide if, when I install Win98 soon, I should go to FAT32.
> So I am inquiring about the concept of 32 bit software and FAT32.
> Do they have anything to do with each other

  We'll come back to this.

> ... and exactly what is meant by 32 bit software? I hear people say
> there is no sense going to FAT32 until more 32 bit software is out.
> And I have heard people say that with 16 bit games they do not run
> well under FAT32. My questions are:  How do you know if you have 16
> bit or 32 bit software, and if you can tell once installed, what
> about before it is bought?

  The Intel 8088, used in the original IBM PC, worked on data 16 bits
at a time.  Every x86-series processor since has included a "real
mode" -- the default when power is first applied -- which behaves the
same as an 8086/8088.  The 286, used in the IBM AT, added an enhanced
addressing capability, but still only worked on 16 bits at a time,
and this "16-bit protected mode" is also included in all later
processors.
  The 386 improved on the 286 in several important ways.  One was
that code could now work on values 32 bits at a time rather than just
16 bits at a time.  Another was the new "32-bit protected mode",
which finally made it easy to write large applications.

  A "32-bit application" is one that runs in "32-bit protected mode",
whether that is provided by Windows or by some other extender such as
DOS/4G.  [You may have bought games that used DOS4GW, a free version
of this extender.]  Specifically, code written for "Win32" should
generally run under Win95, Win98 or NT 4.0, and *may* run under NT
3.5x and/or Win 3.11 with Win32S installed.

  Since there is very little 16-bit software being produced any more,
any advice to wait "until more 32 bit software is out" has been out
of date for about two years -- well before FAT32 was made available.

> My understanding was that FAT 32 had to do with how files are stored on the
> hard drive, namely reducing cluster size to 4K.

  This is correct.  The first DOS hard drives used FAT12, but I think
it was 3.3 that upped this to FAT16.  In each case, the number is the
number of bits used to identify a cluster.  More bits means the
system can handle more clusters -- means each cluster can be smaller.


  There is a special case of "32-bit software" built into Win 3.1x,
which may be contributing to this confusion.  From the "386 Enhanced"
control panel, you could enable two "32-bit access" options.  One was
a replacement for the DOS file-system code, and one for the BIOS drive
access code.  If you had a compatible driver for your controller, you
could speed file/disk I/O by using these instead of the default 16-bit
DOS/BIOS routines.


> Also, I hear that WIN NT is truly a 32 bit software, but it will
> not run under FAT32.

  There are third-party drivers to allow NT to deal with FAT32
partitions, and I believe Microsoft has committed to provide one in
NT 5.0.  NT has its own drive format, NTFS, which is not supported
natively from other versions of Windows -- again, there are some
third-party products available....

David G

             PCSOFT maintains many useful files for download
              on our web site - visit our download page at:
                     http://nospin.com/pc/files.html