What BIOS will show doesn't have much to do with what the BIOS will work with. My 1994 Pentium won't work with any geometry with more than 4096 cylinders, but you can put in any values you want and see the result; it just won't work. Also, CHS geometry only gets you so far. Any drive over 8.4GB or so reports a geometry of 16383/16/63 while actual capacity moves to an IDE bus command. There's also the int 13 limit, etc. etc. Very large drives on very old PCs just doesn't work very well.

If you create a 16Gb partition and put that disk in an old PC and FDISK sees it as a 16GB partition, all that means is that FDISK can read the partition table. When it comes time to create a new partition, it can only do as much as the BIOS and IDE controller allow.

JudgeMonroe wrote:

What BIOS will show doesn't have much to do with what the BIOS will work with. My 1994 Pentium won't work with any geometry with more than 4096 cylinders, but you can put in any values you want and see the result; it just won't work. Also, CHS geometry only gets you so far. Any drive over 8.4GB or so reports a geometry of 16383/16/63 while actual capacity moves to an IDE bus command. There's also the int 13 limit, etc. etc. Very large drives on very old PCs just doesn't work very well.

If you create a 16Gb partition and put that disk in an old PC and FDISK sees it as a 16GB partition, all that means is that FDISK can read the partition table. When it comes time to create a new partition, it can only do as much as the BIOS and IDE controller allow.

Prior to the INT48 limit, these issues were almost always a problem with BIOS or BIOS extension code. That's why XTIDE can usually give access up to the INT48 limit when loaded from a ROM in a NIC card with no hardware modifications needed.

JudgeMonroe wrote:

What BIOS will show doesn't have much to do with what the BIOS will work with. My 1994 Pentium won't work with any geometry with more than 4096 cylinders, but you can put in any values you want and see the result; it just won't work. Also, CHS geometry only gets you so far. Any drive over 8.4GB or so reports a geometry of 16383/16/63 while actual capacity moves to an IDE bus command. There's also the int 13 limit, etc. etc. Very large drives on very old PCs just doesn't work very well.

If you create a 16Gb partition and put that disk in an old PC and FDISK sees it as a 16GB partition, all that means is that FDISK can read the partition table. When it comes time to create a new partition, it can only do as much as the BIOS and IDE controller allow.

Thanks for clarifying some things. I came across a lot of information concerning limitations but with hardware being so diverse, it's hard to know which limitations affect my PC. DOS 7.1 has no size limitation that will be a problem, that I know and if I set cylinders to 4000 and heads to 127 I can easily get 16GB out of the drive according to the BIOS but you're right - DOS needs to be able to correctly "talk" to the BIOS AND address the full disk size. I found a utility which confirms that the BIOS has no Int 13 extensions so 8GB will indeed be the maximum. This means I'll need to manually set the CHS to a setting that gives me 8GB combined with DOS 7.1 which will be 4 times more than what DOS 6.22 can give me + less wasted space due to how FAT32 works.

gdjacobs wrote:

JudgeMonroe wrote:

What BIOS will show doesn't have much to do with what the BIOS will work with. My 1994 Pentium won't work with any geometry with more than 4096 cylinders, but you can put in any values you want and see the result; it just won't work. Also, CHS geometry only gets you so far. Any drive over 8.4GB or so reports a geometry of 16383/16/63 while actual capacity moves to an IDE bus command. There's also the int 13 limit, etc. etc. Very large drives on very old PCs just doesn't work very well.

If you create a 16Gb partition and put that disk in an old PC and FDISK sees it as a 16GB partition, all that means is that FDISK can read the partition table. When it comes time to create a new partition, it can only do as much as the BIOS and IDE controller allow.

Prior to the INT48 limit, these issues were almost always a problem with BIOS or BIOS extension code. That's why XTIDE can usually give access up to the INT48 limit when loaded from a ROM in a NIC card with no hardware modifications needed.

I looked this up half a year ago and it seemed interesting. Oddly enough, none of my PCs (not even my 386) has any problems handling 2GB drives under DOS 6.22 so I never bothered following up. My IBM P166 (upgraded to 233MMX) reads SSDs fine so I have a massive 32GB drive there too. It's only because I'm greedy and never happy 😉 that I wanted to have my HP Pentium 70 to hold nearly my entire 1990-1995 library of games and then FAT16 kind of sucks with a 3MB game using 5MB of disk space. If I get 8GB I'll be happy - should be enough for all the games I need!

red_avatar wrote:

less wasted space due to how FAT32 works.

Not sure what you mean here, FAT32 doesn't support smaller cluster sizes so it would only ever be less efficient.