Storage Unit Converter (GB, GiB, TB, TiB, MB, MiB)

Background.

There are two different meanings of the word gigabyte in daily use, and almost every complaint about a drive being smaller than advertised comes down to a collision between them. To a storage manufacturer a gigabyte is 10⁹ bytes, because in the International System of Units the prefix giga means a thousand million and nothing else. To an operating system a gigabyte is usually 2³⁰ bytes, because computer memory is addressed in powers of two and the industry borrowed the SI prefixes to describe them. The two differ by 7.374%, and the gap grows at every step up the ladder. This converter shows both at once so you can stop guessing which one a tool meant.

NIST puts the split plainly: manufacturers of computer memory use megabyte to mean 2²⁰ = 1,048,576 bytes, while manufacturers of computer storage devices usually use the term to mean 1,000,000 bytes. Neither is wrong; they are simply different conventions applied to different products, and the same vendor will use both in a single document. A Micron DDR4 memory module datasheet describes the part as 8GB — memory, so binary, 8 × 2³⁰ bytes — and quotes its transfer rate on the same line as 25.6 GB/s, which is decimal. That is not carelessness, it is the convention working exactly as documented.

The fix the standards bodies settled on was to give the binary powers their own names. IEC 60027-2 defines kibi (Ki, 2¹⁰), mebi (Mi, 2²⁰), gibi (Gi, 2³⁰), tebi (Ti, 2⁴⁰), pebi (Pi, 2⁵⁰) and exbi (Ei, 2⁶⁰), and NIST's reasoning for adopting them is characteristically blunt: computer professionals needed to talk to physicists, engineers and ordinary people, most of whom know that a kilometre is 1,000 metres. Linux tools, macOS since 10.6 and most storage documentation now use the IEC names correctly. Windows does not — it divides by powers of two and writes GB anyway, which is why a 1 TB drive appears there as roughly 931 GB when what it means is 931 GiB.

The second confusion this page settles is bits against bytes. A lower-case b is a bit and an upper-case B is a byte, and eight bits make one byte, so an 8 Mb file is a 1 MB file and a 100 Mb/s connection moves at most 12.5 MB/s. Network and memory-interface specifications quote bits; file sizes and disk capacities quote bytes. Both are in the dropdowns here, on the same ladder, so a figure copied from a datasheet can be compared directly with one copied from a file manager.

Everything is normalised to a byte count before being divided back out, so any of the sixteen units converts to any other without a special case, and the byte count itself is always shown. If the two numbers on your screen disagree, the byte count is the one to trust — the difference between them is a naming convention, not missing storage. What this page does not do is model filesystem overhead: formatting, reserved blocks, journals and metadata all consume real space on top of the arithmetic here, and how much depends on the filesystem rather than on the units.

What is storage unit converter?

A storage unit converter maps a quantity of digital information between the units used to describe it. Those units come in three families. Bit units count individual binary digits and use decimal SI prefixes: a kilobit is 1,000 bits, a megabit 10⁶, a gigabit 10⁹. Decimal byte units count groups of eight bits and use the same SI prefixes: a kilobyte is 10³ bytes, a megabyte 10⁶, a gigabyte 10⁹, a terabyte 10¹², a petabyte 10¹⁵. Binary byte units use the IEC prefixes standardised in IEC 60027-2 and count powers of two: a kibibyte is 2¹⁰ = 1,024 bytes, a mebibyte 2²⁰, a gibibyte 2³⁰, a tebibyte 2⁴⁰, a pebibyte 2⁵⁰. Because both families borrow the same spoken names — 'gigabyte' for both 10⁹ and 2³⁰ — the same figure can be written two ways that differ by an amount that grows with the prefix: 2.4% at the kilo tier, 4.858% at mega, 7.374% at giga, 9.951% at tera and 12.590% at peta. Converting between them is exact arithmetic on a byte count, with no approximation anywhere, which is why this page always reports the byte count alongside the answer.

How to use this calculator.

1. 01Enter the amount and pick the unit it is currently written in. If you copied the figure off a drive, a network plan or an invoice it is almost certainly a decimal unit; if you copied it out of an operating system, a memory datasheet or a RAID controller it is almost certainly a binary one.
2. 02Pick the unit you want it in. Any of the sixteen converts to any other, including bits to pebibytes, because everything routes through a byte count.
3. 03Read the byte count as the arbiter. When two tools disagree about how big something is, converting both to bytes usually shows they agree completely and are just labelling the answer differently.
4. 04Use the ladder below the main result to see the same quantity in MB, MiB, GB, GiB, TB and TiB at once. That is the fastest way to spot which convention a piece of software is using.
5. 05Watch the case of the b. Mb is megabits and MB is megabytes, a factor of eight apart, and mixing them up is the single most common error when sizing a transfer against a link speed.

The formula.

Every conversion is two multiplications. First the input is turned into bytes by multiplying by the number of bytes in its unit; then that byte count is divided by the number of bytes in the target unit. There is no rounding in between, and no unit pair is special-cased, so bits-to-pebibytes goes through exactly the same path as megabytes-to-mebibytes.

The byte counts themselves come from two standards. The decimal prefixes are the SI prefixes defined by the BIPM: kilo is 10³, mega 10⁶, giga 10⁹, tera 10¹², peta 10¹⁵. The binary prefixes are those of IEC 60027-2, listed by NIST as kibi 2¹⁰, mebi 2²⁰, gibi 2³⁰, tebi 2⁴⁰, pebi 2⁵⁰ and exbi 2⁶⁰. Bit units use the decimal prefixes and are then divided by eight to reach bytes, so one megabit is 10⁶ ÷ 8 = 125,000 bytes.

The gap between a decimal unit and its binary neighbour is just the ratio of the two definitions, and because each binary step multiplies by 1,024 while each decimal step multiplies by 1,000, that ratio compounds. 2¹⁰ ÷ 10³ = 1.024, so a kibibyte is 2.4% larger than a kilobyte. At the mega tier it is 1.024² = 1.048576, or 4.858%. At giga, 1.024³ = 1.073741824, or 7.374%. At tera, 1.024⁴, or 9.951%. At peta, 1.024⁵, or 12.590%. The pattern is the reason the complaint gets louder as drives get bigger: nobody noticed on a 1.44 MB floppy, everybody notices on a 20 TB drive that reports 18.2.

Working an example the other way makes it concrete. A gibibyte is 2³⁰ = 1,073,741,824 bytes, so 1 GiB is 1.073741824 GB. Turn that around and 1 GB is 1 ÷ 1.073741824 = 0.9313225746 GiB. Multiply by a thousand and a 1 TB drive is 931.3225746155 GiB — which Windows will display as '931 GB'. The drive holds precisely the trillion bytes it was sold as; only the divisor changed.

One last note on precision. All the factors here are exact integers, and the arithmetic is carried at Decimal precision and rounded once at the end to ten decimal places. That is enough to keep every conversion between the units on this page exact to well beyond any practical need, but it does mean a genuinely microscopic result — one bit expressed in pebibytes, for instance — displays as zero rather than as 1.1 × 10⁻¹⁶. The byte count is always available if you need the exact figure.

A worked example.

Frequently asked questions.

References& sources.

1. [1]National Institute of Standards and Technology. Prefixes for binary multiples (Reference on Constants, Units and Uncertainty), citing IEC 60027-2 second edition, 2000-11. Source of the binary prefixes used here — kibi Ki 2¹⁰, mebi Mi 2²⁰, gibi Gi 2³⁰, tebi Ti 2⁴⁰, pebi Pi 2⁵⁰, exbi Ei 2⁶⁰ — and of the statement quoted on this page that most memory manufacturers use megabyte to mean 2²⁰ = 1,048,576 bytes while manufacturers of computer storage devices usually use it to mean 1,000,000 bytes.
2. [2]Bureau International des Poids et Mesures. SI prefixes. Source of the decimal prefixes: kilo 10³, mega 10⁶, giga 10⁹, tera 10¹², peta 10¹⁵ and exa 10¹⁸, described as the prefixes for writing decimal multiples and submultiples of SI units. Also records the 2022 extension of the prefix range to ronna and quetta.
3. [3]Micron Technology. MTA8ATF1G64HZ 8GB DDR4 SDRAM SODIMM datasheet. Cited as the concrete example of both conventions appearing in one vendor document: the module capacity is given as 8GB (memory, binary — 8 × 2³⁰ bytes) while its transfer rate on the same line is given as 25.6 GB/s (decimal — 3200 MT/s × 8 bytes ÷ 1000).