| Exabits per second (Ebit/s) | Tebioctets per second (Tio/s) |
|---|---|
| 1 Exabit per second | 113686.837722 Tio/s |
| 2 Exabits per second | 227373.675443 Tio/s |
| 3 Exabits per second | 341060.513165 Tio/s |
| 4 Exabits per second | 454747.350886 Tio/s |
| 5 Exabits per second | 568434.188608 Tio/s |
| 10 Exabits per second | 1136868.37722 Tio/s |
| 20 Exabits per second | 2273736.75443 Tio/s |
| 25 Exabits per second | 2842170.94304 Tio/s |
| 50 Exabits per second | 5684341.88608 Tio/s |
| 100 Exabits per second | 11368683.7722 Tio/s |
| Reference | Exabits per second (Ebit/s) | Tebioctets per second (Tio/s) |
|---|---|---|
| A dial-up modem | 5.6 × 10-14 Ebit/s | 0.00000000636646 Tio/s |
| Typical home broadband | 1 × 10-10 Ebit/s | 0.0000113687 Tio/s |
| Gigabit Ethernet | 0.000000001 Ebit/s | 0.000113687 Tio/s |
| Streaming a 4K film | 2.5 × 10-11 Ebit/s | 0.00000284217 Tio/s |
The exabit per second is a unit of data transfer rate equal to a thousand petabits per second. Its symbol is Ebit/s. No link, no cable and no exchange point runs at this rate; the unit describes the internet as a whole, or a large part of it, at a single instant.
Global internet traffic can be expressed this way. Worldwide traffic of several hundred exaoctets a month works out to roughly one exabit per second on average, and peak hours run higher. That figure is the sum of every packet moving on every network on the planet, and it is the only quantity anyone routinely describes at this scale.
An exabit per second is 125 petaoctets per second. Nothing generates data at that rate in one place; the number is an aggregate of billions of separate flows, most of them tiny. A single video stream is a few megabits per second, so an exabit per second is on the order of a hundred million such streams running at once.
The composition of that traffic is dominated by video. Streaming services account for the largest share, followed by social platforms, software updates and cloud synchronisation. Ordinary web browsing and messaging, which people think of as the internet, are a small fraction of the total by volume even though they occupy most of the attention.
The unit also appears in projections of aggregate capacity. The total installed capacity of all submarine cables, if every wavelength on every fibre pair were lit and used simultaneously, is in the exabit-per-second range. Actual utilisation is far below that, because capacity is built ahead of demand and because routes must carry each other's traffic when a cable fails.
The prefix exa is a thousand to the sixth power. It was adopted in 1975, when nobody expected it to describe anything but astronomical quantities, and it now describes the working throughput of a communications system built by human beings. The interval between definition and everyday use was about forty years.
One exabit per second equals 1,000 petabits per second, 125 petaoctets per second, or about 0.8674 exbibits 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.