| Mebibits (Mibit) | Exabits (Ebit) |
|---|---|
| 1 Mebibit | 1.048576 × 10-12 Ebit |
| 2 Mebibits | 2.097152 × 10-12 Ebit |
| 3 Mebibits | 3.145728 × 10-12 Ebit |
| 4 Mebibits | 4.194304 × 10-12 Ebit |
| 5 Mebibits | 5.24288 × 10-12 Ebit |
| 10 Mebibits | 1.048576 × 10-11 Ebit |
| 20 Mebibits | 2.097152 × 10-11 Ebit |
| 25 Mebibits | 2.62144 × 10-11 Ebit |
| 50 Mebibits | 5.24288 × 10-11 Ebit |
| 100 Mebibits | 1.048576 × 10-10 Ebit |
| Reference | Mebibits (Mibit) | Exabits (Ebit) |
|---|---|---|
| A plain text message (160 characters) | 0.0012207 Mibit | 1.28 × 10-15 Ebit |
| A three-minute MP3 | 22.8882 Mibit | 2.4 × 10-11 Ebit |
| A smartphone photo | 30.5176 Mibit | 3.2 × 10-11 Ebit |
| A high-definition film | 30517.6 Mibit | 0.000000032 Ebit |
| A dual-layer Blu-ray disc | 381470 Mibit | 0.0000004 Ebit |
The mebibit is a unit of digital information equal to 1,048,576 bits, which is 1,024 kibibits or two to the twentieth power. Its symbol is Mibit. It is the binary counterpart of the megabit, and the difference between the two has grown to 4.9 per cent — small enough to be ignored casually, large enough to matter in a specification.
Its natural home is semiconductor memory, where capacity is fixed by the address lines on the chip. A memory die with twenty address bits can reach exactly 1,048,576 locations, so densities land on powers of two rather than on round decimal numbers. A chip described as 512 megabits in a catalogue is very often 512 mebibits in reality, which is 537 megabits, and the datasheet is where the truth is written.
That habit is old and deliberate. Semiconductor manufacturers have always sized memory in binary because the wiring makes any other choice wasteful: an address bus is a fixed number of lines, and using only part of the range it can reach throws away silicon. Storage manufacturers had no such constraint, which is precisely why the two industries diverged.
For a sense of quantity, a mebibit is 131,072 octets, or about 128 kibioctets. That is roughly the contents of a long book as plain text, or a single low-resolution image. In network terms it is a hundredth of a second of a gigabit link.
The IEC prefixes were introduced in 1998 partly because at this level and above the approximation stops being safe. Buying a component described as one megabit and receiving one mebibit is a five per cent surprise; at the tebibit the same confusion is a ten per cent surprise, and in a contract for a data centre that is a real sum of money.
Correct usage is now common in standards documents, in hardware datasheets and in operating system internals, and rare in advertising and consumer software. The best rule when reading a figure is to ask which industry produced it: if it describes memory or a chip, assume binary; if it describes storage or a link, assume decimal.
One mebibit equals 1,048,576 bits, 1,024 kibibits, 131,072 octets, or about 1.049 megabits.
The exabit is a unit of digital information equal to a thousand petabits, or a billion billion bits. Its symbol is Ebit. At this scale no device, no cable and no building is being described. The exabit measures the traffic of whole regions and the output of entire industries over stretches of time.
An exabit is 125 petaoctets. The most common place to meet the unit is in figures for global internet traffic, which are usually given in exaoctets per month. Worldwide traffic passed one exaoctet per month around 2004, reached about a hundred by 2017 and is now several hundred, which in bits is several thousand exabits every month.
Those totals divide in revealing ways. The overwhelming majority is video, and the rest is dominated by software updates, cloud synchronisation and machine-to-machine traffic rather than by anything a person is actively reading. A single popular film released simultaneously worldwide can move a measurable fraction of an exabit in a day.
Storage at this scale exists only as aggregates. The combined capacity of all the hard drives and flash memory manufactured in a year is measured in hundreds of exaoctets, which is thousands of exabits. Estimates of all the data held by humanity at any moment run into the tens of zettaoctets, and the exabit is the unit one step below that in which the components of such estimates are counted.
Scientific instruments produce exabit quantities before any of it is stored. The Square Kilometre Array radio telescope generates raw data at a rate that would fill exaoctets within days, and its design is built around discarding and summarising in real time rather than recording. The same is true of the large particle detectors and of high-resolution climate simulations.
The prefix exa comes from the Greek for six, the exabit being a thousand to the sixth power in bits. It was adopted in 1975 along with peta, at a time when no one expected either to describe anything but astronomical or theoretical quantities. Both are now ordinary units in the technology industry, which says something about how fast the numbers grew.
One exabit equals 1,000 petabits, 125 petaoctets, or about 0.8674 exbibits.