And some addition after question from AI "What about DOS?":
Windows 98 SE, DOS VM, COMMAND.COM, DIR /V

1 Volume in drive D has no label
2 Volume Serial Number is FC4B-3417
3 Directory of D:\
4File Name         Size        Allocated      Modified      Accessed  Attrib
5
6..             <DIR>                                                     D      ..
7         0 file(s)              0 bytes
8         1 dir(s)               0 bytes allocated
9                   133 168 812.00 MB free
10                   133 169 151.94 MB total disk space,   0% in use

Three decades aged code does not have any problem to the credit of its developers.

SweetLow wrote on 2026-09-01, 07:54:

There are few topics on this theme, like this previous one What is the biggest storage HDD Windows 98SE can use? But what are th […]
Show full quote

There are few topics on this theme, like this previous one What is the biggest storage HDD Windows 98SE can use?
But what are the limits you can get really?
Ok, lets go through the storage levels.
1. IOS level. Drives and partitions. 64-bit LBA supported, so the only limits are inside drivers for hardware. As there are 48-bit LBA for ATA drives and 64-bit LBA for SCSI drives and GPT support available now this level does not have any real restrictions except physical or emulated drive size.
2. IFSMGR volume level. File systems and they drivers. IFSMGR limit itself is 47-bit LBA for one volume but it is structured limit: 32 bit (4G) for cluster (allocation unit) count and 15 bit (32K) for cluster size measured in sectors. And real available file system driver for NTFS (Paragon UFSD) has its own limit: 1G clusters * 64KB cluster size on original version and 2G clusters on patched version. So the real available volume size is slightly less than 128TB now.
3. IFSMGR file level. File systems and they drivers. IFSMGR limit itself is 32-bit bytes for file and file system drivers have this limit too.

Ok. And some proof, of course - Windows 98 SE + patched Paragon UFSD @ QEMU:

The attachment 127T_ENG.png is no longer available

I bought a USB HDD off ebay a little while ago, it was stupid cheap for a 4Tb disk. But I went for it anyway.
When I got it I formatted it and it was fine, and copied a load of data to it no problem.

Then I come to try and read that data and it wasnt really there.
It was a 20Gb HDD with some firmware tricks to make it report it was 4Tb.
I wonder what Win98 would make of a drive like that. You could set the flags to be 400Tb right?

ElectroSoldier wrote on 2026-09-21, 19:24:

You could set the flags to be 400Tb right?

Not flag of course, but with help of driver you can really set any size in internal structure (DCB) representing drive for OS in protected mode.

The question I suppose is: who isn't to say that the moment you try to write more than two 2 TB it won't wraparound and destroy all data? A number of "limits" manifest that way...
(i.e. is the displayed free space really indicative of anything?)

Three decades aged code does not have any problem to the credit of its developers.

There's about nothing of the original three decades old code running in such heavily patched setup 😀 other than maybe printing a 64-bit integer?

javispedro1 wrote on 2026-09-22, 12:54:

The point, I suppose, he's making is: who isn't to say that the moment you try to write more than two 2 TB it won't wraparound and destroy all data?

You asked the right question but too late as it answered already in 2023 when GPTTSD was tested after its release exactly for handling of more than 2TB volumes in UFSD/PNTFS.

There's about nothing of the original three decades old code running in such heavily patched setup 😀 other than maybe printing a 64-bit integer?

Of course not only printing (and it is not usual for DOS code to do the right 64-bit math or even ANY 64-bit math, I can say). The code that maps PM structure to DOS INT works as expected too.