Why does Windows always reports slightly wrong sizes when formatting a disk?

Question

I've installed Windows probably thousands of times since Windows 3.1. I forget the situation in Win 3.1, but from Windows 95 onwards, Windows has never respected partition size reality, and as it's happening again today, I thought maybe someone on here could enlighten me...

e.g. when I have a newly wiped hard disk, Windows asks me what size I would like the new system partition to be, in MB, so, as I want 64 GB in this case, I put in 65536 ... exactly 64 GB, right? Wrong! ... as soon as you do that, Windows reports the new partition size as 63.9 GB.

Have Microsoft been playing a 20+ year practical joke on us?? ... or, is there a good reason why Microsoft report 65536 MB in Windows as being 63.9 GB instead of 64 GB?

This is not the most important question to know, but I'd love to know if someone has a good answer as it's made me scratch my head for a long time (and if this can be explained by NTFS tables taking a bit of space or something, what would you say is the correct way to get Windows to setup a 64 GB drive)? 🙂

Answer 1

Your trial and error method is a good idea but instead of looking at places that only show you the leading digits of a lengthy figure you should rather look at the properties of your partition/volume where you can find an exact length in bytes. That information is more helpful for your method of trying to approach the target size in an iterative matter.

When using Windows 7 (64 bit) I get shown the exact byte size and an abbreviated number based on leading digits which seems to correspond to the numbers shown in Windows explorer. Obviously, the abbreviated number that you were looking at is not a good measure to determine the real length!

Why trial and error?

The reason to run a trial and error strategy is that Windows formatting behaviour might have changed over time.

Disk geometry influenced partitioning first

Initially, influenced by the old CHS (cylinder, head, sector) notion in the MBR partition table, Windows XP always tried to place a partition start at the beginning of a cylinder (x,0,1) if I remember correctly.

Advanced format influences partitioning now

With the arrival of a genuine physical sector size of 4096 bytes (advanced format) and the emulation of 512 bytes logically (at the interface) the old formatting policy leads to misalignment that results in partitions starting somewhere at a multiple of 512 byte which is always at the beginning of a logical sector but not necessarily at the beginning of a physical sector with a size of 4096 bytes. Such misalignment with regards to the physical sector size is causing avoidable wear and slowdown when writing. Later Microsoft operating systems must have allocated partition space differently to avoid such a situation.

The Gparted partitioning tool in some linux operating systems has the option to allocate space in multiple of a MB (don't know if that was a flat 10^6 figure or 2^20 bytes). Maybe using a pendrive linux helps you to match your formatting goals.

Please read the following article to be aware of the mess that comes with size terms:

https://en.wikipedia.org/wiki/Binary_prefix#gibi