Why Storage Units Never Seem to Add Up

Every year millions of people buy a 1 terabyte drive, plug it in, and find that the computer reports 931 gigabytes. Nothing has been lost and nothing was misadvertised in a legal sense — two different definitions of the same word are colliding. Drive manufacturers count in powers of ten, where a terabyte is 1,000,000,000,000 bytes. Operating systems count in powers of two, where a terabyte is 1,099,511,627,776 bytes. The drive holds exactly the number of bytes printed on the box; Windows simply divides that count by a larger denominator, and the quotient looks smaller.

This converter shows both columns at once so you never have to guess which convention a number came from. Enter any capacity in bits, bytes, kilobytes, megabytes, gigabytes, terabytes or petabytes and read across to every other unit under both standards, with the exact byte count as the anchor.

Binary and Decimal Prefixes Explained

Computers address memory in powers of two, so early programmers borrowed the metric prefix "kilo" for 1,024 bytes because it was close enough to a thousand. The approximation held up at small scales but compounds badly: the error is 2.4% at kilobytes, 4.9% at megabytes, 7.4% at gigabytes, 10.0% at terabytes and 12.6% at petabytes. That growing wedge is precisely the gap you see between the box and the drive properties dialog.

In 1998 the International Electrotechnical Commission standardised separate names: kibibyte (KiB), mebibyte (MiB), gibibyte (GiB) and tebibyte (TiB) for the binary values, leaving KB, MB, GB and TB to the decimal ones. Linux tooling and macOS adopted the split, and macOS now reports a 1 TB drive as a full 1 TB. Windows kept binary arithmetic while continuing to print the decimal abbreviations, which is the single largest reason the confusion survives. RAM is always quoted in binary units regardless of platform, because memory chips really are manufactured in powers of two.

Raw Capacity Versus Usable Capacity

Unit conversion is only half the story. A freshly formatted volume gives back less than its raw size because the filesystem needs room for its own bookkeeping: the master file table or inode tables, journals, superblock copies, and allocation bitmaps. NTFS and APFS typically consume a few tenths of a percent up to about one percent, while ext4 reserves five percent for the root user by default so that a full disk does not lock an administrator out. Vendor recovery partitions take a further fixed slice, often 10 to 30 gigabytes on a prebuilt laptop.

Small files pay an extra tax through internal fragmentation. Storage is allocated in clusters, so a 2 kilobyte text file occupies a full 4 kilobyte cluster and wastes the remainder. A library of many tiny files therefore consumes noticeably more space than the sum of its file sizes suggests. The Advanced Options let you model overhead and reserved space so the usable figure reflects the disk you actually have rather than the one on the label.

Planning Transfers, Backups and Capacity

Once you know the usable capacity, two practical questions usually follow: how long will it take to move, and how much room do the copies need. Transfer time trips people up because network and interface speeds are quoted in bits per second while files are measured in bytes. Eight bits to the byte means a 100 Mbps link moves at best about 12.5 MB per second, and real throughput lands lower after protocol overhead. Professional Mode takes megabytes per second directly so the arithmetic stays honest.

For backup planning, the widely used 3-2-1 rule calls for three copies of the data on two kinds of media with one kept offsite, which means budgeting roughly three times the source size before any deduplication or compression. Set the number of copies to see the full footprint. Between the unit table, the usable-capacity model and the transfer estimate, you can size a drive, a backup target or a migration window without converting anything by hand.