Chuck(G)
25k Member
Woah...
We're talking about two different notions of "32 bit".
Windows 95/98 and ME are largely 32-bit code. That is, they use code that makes use of 32-bit registers and addressing on an 80386 or above. In a sense, it's a whole different instruction set. VxDs, for example, are 32-bit executable code. Even Windows 3.0 can make use of 32-bit code. There are even DOS programs that, with the proper "extender" can use 32-bit code. An 80286 cannot run 32-bit code.
That's a whole different issue from FAT12 vs. FAT16 vs. FAT32. The names refer to the size of the bit field used to link disk blocks or clusters. Originally, DOS started with 12-bit cluster enumeration--and that's what's used even today for most floppies. That is, the disk is divided into up to 4,090 or so "clumps" or "clusters" of sectors. On a 360K drive, this is quite natural, as there are only 720 sectors to link together. Initially, hard disks started out this way. but that can be very limiting, both in the amount of storage than can be so enumerated and in the number of files that can be stored.
Let me detour a bit and explain the FAT (File Allocation Table). It's nothing more than a list of clusters (groups of sectors) that's stored in a predefined location on a disk partition or drive. Each slot in this list either points to nothing (the cluster is available for allocation) or signifies the last cluster in a chain of clusters for a file, or points to the next cluster in the chain of clusters that describes the storage used for a file. Just that simple--a FAT is a list. Without it, you've got no way to determine if clusters of sectors are in use or even what their place in a file is.
When the 12-bit FAT became inadequate as hard disks became larger, a 16-bit slot was adopted for hard disks, so up to roughly 65,000 clusters of sectors can be described. Eventually (by about DOS 6), even this was inadequate, so a 32-bit slot was adapted, allowing (theoretically) about 4 billion clusters to be described (it's actually much less, as certain values in the 32-bit slot are reserved for special uses)--and this is what's used for larger partitions in Win9x. (FDISK selects the type of FAT automatically based on the size of the partition, so, for, say a 100MB partition, a 16-bit FAT type will be used).
But 32-bit FAT file systems can be accessed just fine in real-mode 16-bit code. FreeDOS, as well as IBM PC-DOS 7.x can do this as can the DOS boot from Win9x (95 OSR2 and beyond).
There's lots on the web that explains it better than I can.
We're talking about two different notions of "32 bit".
Windows 95/98 and ME are largely 32-bit code. That is, they use code that makes use of 32-bit registers and addressing on an 80386 or above. In a sense, it's a whole different instruction set. VxDs, for example, are 32-bit executable code. Even Windows 3.0 can make use of 32-bit code. There are even DOS programs that, with the proper "extender" can use 32-bit code. An 80286 cannot run 32-bit code.
That's a whole different issue from FAT12 vs. FAT16 vs. FAT32. The names refer to the size of the bit field used to link disk blocks or clusters. Originally, DOS started with 12-bit cluster enumeration--and that's what's used even today for most floppies. That is, the disk is divided into up to 4,090 or so "clumps" or "clusters" of sectors. On a 360K drive, this is quite natural, as there are only 720 sectors to link together. Initially, hard disks started out this way. but that can be very limiting, both in the amount of storage than can be so enumerated and in the number of files that can be stored.
Let me detour a bit and explain the FAT (File Allocation Table). It's nothing more than a list of clusters (groups of sectors) that's stored in a predefined location on a disk partition or drive. Each slot in this list either points to nothing (the cluster is available for allocation) or signifies the last cluster in a chain of clusters for a file, or points to the next cluster in the chain of clusters that describes the storage used for a file. Just that simple--a FAT is a list. Without it, you've got no way to determine if clusters of sectors are in use or even what their place in a file is.
When the 12-bit FAT became inadequate as hard disks became larger, a 16-bit slot was adopted for hard disks, so up to roughly 65,000 clusters of sectors can be described. Eventually (by about DOS 6), even this was inadequate, so a 32-bit slot was adapted, allowing (theoretically) about 4 billion clusters to be described (it's actually much less, as certain values in the 32-bit slot are reserved for special uses)--and this is what's used for larger partitions in Win9x. (FDISK selects the type of FAT automatically based on the size of the partition, so, for, say a 100MB partition, a 16-bit FAT type will be used).
But 32-bit FAT file systems can be accessed just fine in real-mode 16-bit code. FreeDOS, as well as IBM PC-DOS 7.x can do this as can the DOS boot from Win9x (95 OSR2 and beyond).
There's lots on the web that explains it better than I can.
