| Yobioctets per second (Yio/s) | Tebioctets per second (Tio/s) |
|---|---|
| 1 Yobioctet per second | 1099511627776 Tio/s |
| 2 Yobioctets per second | 2199023255552 Tio/s |
| 3 Yobioctets per second | 3298534883328 Tio/s |
| 4 Yobioctets per second | 4398046511104 Tio/s |
| 5 Yobioctets per second | 5497558138880 Tio/s |
| 10 Yobioctets per second | 10995116277760 Tio/s |
| 20 Yobioctets per second | 21990232555520 Tio/s |
| 25 Yobioctets per second | 27487790694400 Tio/s |
| 50 Yobioctets per second | 54975581388800 Tio/s |
| 100 Yobioctets per second | 109951162777600 Tio/s |
| Reference | Yobioctets per second (Yio/s) | Tebioctets per second (Tio/s) |
|---|---|---|
| A dial-up modem | 5.79026 × 10-21 Yio/s | 0.00000000636646 Tio/s |
| Typical home broadband | 1.03398 × 10-17 Yio/s | 0.0000113687 Tio/s |
| Gigabit Ethernet | 1.03398 × 10-16 Yio/s | 0.000113687 Tio/s |
| Streaming a 4K film | 2.58494 × 10-18 Yio/s | 0.00000284217 Tio/s |
The yobioctet per second is a unit of data transfer rate equal to 1,024 zebioctets per second, or two to the eightieth power octets per second. Its symbol is Yio/s. It is the largest binary rate the International Electrotechnical Commission has named, and the last rung of the ladder that began with the bit per second.
At this final step the binary and decimal conventions differ by 20.9 per cent. That figure closes the argument the IEC prefixes were created to settle: a naming habit that was 2.4 per cent wrong at the kibibit per second has grown, through eight successive multiplications by 1.024, into a discrepancy of more than a fifth. No system of measurement can carry an ambiguity that large, and the whole binary series exists to remove it.
The rate itself has no referent. A yobioctet per second is more than eight million times the total instantaneous traffic of the internet, and it would transfer everything humanity has ever stored several thousand times over in a second. Nothing produces data at that rate, nothing consumes it, and nothing is designed with it in view.
That does not make the unit pointless. Defining the whole ladder in advance means that the rule — every prefix combines with every unit — is all anyone has to learn, and a rule is easier to carry than a table of exceptions. The same principle gave the metric system its complete prefix set, and the addition of ronna and quetta in 2022 extended the decimal side without any matching binary names being defined.
The practical value of a unit like this is that it makes a converter's behaviour uniform. A tool that handles the impossible cases by the same rule as the ordinary ones can be trusted not to have special cases hidden in it, and that is a property worth having in something whose whole purpose is to be relied upon.
For the reader, the whole series comes down to one character: Kio/s, Mio/s, Gio/s, Tio/s, Pio/s, Eio/s, Zio/s and Yio/s are binary; ko/s, Mo/s, Go/s, To/s, Po/s, Eo/s, Zo/s and Yo/s are decimal; and the difference between them widens from a rounding error to a fifth as you climb.
One yobioctet per second equals 1,024 zebioctets per second, 1,208,925,819,614,629,174,706,176 octets per second, or about 1.209 yottaoctets per second.
The tebioctet per second is a unit of data transfer rate equal to 1,024 gibioctets per second, or two to the fortieth power octets per second. Its symbol is Tio/s. It is the binary counterpart of the teraoctet per second, and the two differ by 10 per cent.
Nothing outside a large machine moves data this quickly. The rate describes the memory bandwidth of an accelerator with stacked memory, the internal fabric of a high-end processor package, or the aggregate throughput of a parallel filesystem spread across thousands of drives. All of these are built from binary structures, and their totals are binary quantities divided by time.
High-performance computing is where the unit is written most often. A supercomputer's storage system is specified by how many tebioctets per second it can deliver to the compute nodes, because that number determines how quickly a simulation can save its state and resume. A machine that computes quickly but writes slowly spends its time waiting.
The ten per cent difference from the decimal unit is significant in that context. A filesystem procured to deliver 10 teraoctets per second and one delivering 10 tebioctets per second differ by a whole teraoctet per second, which in a facility of that size represents a substantial fraction of the hardware budget.
In bits a tebioctet per second is 8 tebibits per second, and in decimal terms about 1.1 teraoctets per second. Expressing the same rate four different ways is routine at this level, because the storage industry, the memory industry, the network industry and the standards bodies each prefer a different one.
For everyday comparison, a tebioctet per second would fill a large consumer hard drive in about twenty seconds. No external interface carries this; the figure describes movement between components inside a single system, where the wires are short and there are very many of them running in parallel.
Graphics processors have brought the rate within reach of a single component. A stack of high-bandwidth memory bonded directly to the processor die delivers well over a tebioctet per second to the chip that uses it, and a card carrying several such stacks passes a few. That bandwidth, rather than raw arithmetic speed, is what limits the training of large models: the arithmetic units sit idle unless the memory can keep them fed. The same reasoning explains why supercomputer designers spend as much effort on the paths between memory and processor as on the processors themselves, and why the rate is quoted in binary units when the memory it describes is addressed in powers of two.
One tebioctet per second equals 1,024 gibioctets per second, 1,099,511,627,776 octets per second, or about 1.100 teraoctets per second.