| Tebioctets per second (Tio/s) | Terabits per second (Tbit/s) |
|---|---|
| 1 Tebioctet per second | 8.79609302221 Tbit/s |
| 2 Tebioctets per second | 17.5921860444 Tbit/s |
| 3 Tebioctets per second | 26.3882790666 Tbit/s |
| 4 Tebioctets per second | 35.1843720888 Tbit/s |
| 5 Tebioctets per second | 43.980465111 Tbit/s |
| 10 Tebioctets per second | 87.9609302221 Tbit/s |
| 20 Tebioctets per second | 175.921860444 Tbit/s |
| 25 Tebioctets per second | 219.902325555 Tbit/s |
| 50 Tebioctets per second | 439.80465111 Tbit/s |
| 100 Tebioctets per second | 879.609302221 Tbit/s |
| Reference | Tebioctets per second (Tio/s) | Terabits per second (Tbit/s) |
|---|---|---|
| A dial-up modem | 0.00000000636646 Tio/s | 0.000000056 Tbit/s |
| Typical home broadband | 0.0000113687 Tio/s | 0.0001 Tbit/s |
| Gigabit Ethernet | 0.000113687 Tio/s | 0.001 Tbit/s |
| Streaming a 4K film | 0.00000284217 Tio/s | 0.000025 Tbit/s |
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.
The terabit per second is a unit of data transfer rate equal to a thousand gigabits per second. Its symbol is Tbit/s. It is the unit of the internet's backbone: the submarine cables, the exchange points and the long-haul optical links that carry traffic between continents.
A single modern transoceanic cable carries several hundred terabits per second. It achieves this not with one enormous channel but with wavelength division multiplexing, which sends dozens of separate colours of light down each fibre at once, and with several fibre pairs in the same cable. Each wavelength carries a few hundred gigabits, and the totals add up.
The historical comparison is worth stating plainly. The first transatlantic telephone cable, laid in 1956, carried thirty-six simultaneous voice calls. A cable laid in the 2020s carries hundreds of terabits per second, enough for hundreds of millions of simultaneous calls. That is a factor of roughly ten million in seventy years, and it was achieved almost entirely by changing what is sent down the glass rather than by laying more cable.
Internet exchange points, where networks meet and hand traffic to one another, publish their throughput in terabits per second. The largest in Europe and Asia peak in the tens of terabits, and those public graphs are among the most reliable measurements of how heavily the internet is being used at a given moment, because they count real traffic rather than capacity.
In octets, a terabit per second is 125 gigaoctets per second — the contents of a large laptop's disc moved every second, continuously. No single storage system can feed such a link; the traffic on these routes is the aggregate of millions of separate connections, each of them tiny by comparison.
Laboratory records go far higher, into petabits per second, using multi-core fibre and hundreds of wavelengths at once, though over short distances under controlled conditions. The gap between what is demonstrated in a laboratory and what is deployed under an ocean has historically been about a decade.
One terabit per second equals 1,000 gigabits per second, 125 gigaoctets per second, or about 0.9095 tebibits per second.