| Exaoctets per second (Eo/s) | Teraoctets per second (To/s) |
|---|---|
| 1 Exaoctet per second | 1000000 To/s |
| 2 Exaoctets per second | 2000000 To/s |
| 3 Exaoctets per second | 3000000 To/s |
| 4 Exaoctets per second | 4000000 To/s |
| 5 Exaoctets per second | 5000000 To/s |
| 10 Exaoctets per second | 10000000 To/s |
| 20 Exaoctets per second | 20000000 To/s |
| 25 Exaoctets per second | 25000000 To/s |
| 50 Exaoctets per second | 50000000 To/s |
| 100 Exaoctets per second | 100000000 To/s |
| Reference | Exaoctets per second (Eo/s) | Teraoctets per second (To/s) |
|---|---|---|
| A dial-up modem | 7 × 10-15 Eo/s | 0.000000007 To/s |
| Typical home broadband | 1.25 × 10-11 Eo/s | 0.0000125 To/s |
| Gigabit Ethernet | 1.25 × 10-10 Eo/s | 0.000125 To/s |
| Streaming a 4K film | 3.125 × 10-12 Eo/s | 0.000003125 To/s |
The exaoctet per second is a unit of data transfer rate equal to a thousand petaoctets per second, or eight exabits per second. Its symbol is Eo/s. Nothing built runs at this rate: it is about eight times the total traffic of the entire internet, counted across every network on the planet at once.
The unit is useful mainly for thought experiments about limits. If every hard drive and flash chip manufactured in a year were read simultaneously at full speed, the combined rate would be in this range. So would the total output of every camera sensor in every phone on Earth if they all recorded at once. These are sums over the whole world's hardware, not rates any system experiences.
There is one situation in which very large data volumes really do move faster than any network, and it puts the unit in perspective. Physically shipping a container of hard drives across an ocean transfers more data per second, averaged over the journey, than any cable. A shipping container holding a few exaoctets crossing the Atlantic in a week works out to several gigaoctets per second, and a truck of drives driven across a city beats almost any local link.
That calculation is not a joke; cloud providers offer it as a service. When a customer needs to move petaoctets into a data centre, the provider ships a lorry full of storage rather than attempting the transfer over a network, because the network would take months. The bandwidth of a vehicle is enormous, though its latency is measured in days.
For the unit itself, an exaoctet per second is 125 petaoctets per second, and it would transfer the world's entire stock of stored data — a few hundred zettaoctets — in about a week of continuous running. No mechanism exists to feed such a link, and none is being designed.
A converter must nevertheless handle the unit, because it appears in aggregate capacity models, in academic papers on the theoretical limits of communication, and in any table that lists the metric prefixes completely. A quantity does not need a use for its name to be well formed.
One exaoctet per second equals 1,000 petaoctets per second, 8 exabits per second, or about 0.8674 exbioctets per second.
The teraoctet per second is a unit of data transfer rate equal to a thousand gigaoctets per second, or eight terabits per second. Its symbol is To/s. It describes memory bandwidth inside the fastest processors and the aggregate throughput of large computing systems, rather than any link between separate machines.
Graphics and accelerator chips are the clearest example. A modern accelerator uses stacked memory placed alongside the processor on the same package, and the bandwidth between them reaches several teraoctets per second. That figure is what allows thousands of arithmetic units to be kept busy at once, and it is now the property that most often determines how fast a machine learning workload runs.
The reason bandwidth rather than arithmetic has become the limiting factor is worth stating. Processors have grown far faster at computing than memory has at supplying data, so a modern chip can perform tens of operations for every octet it reads. Any calculation that touches memory more often than that is limited by the memory, and most real calculations are.
Supercomputer interconnects reach this range in aggregate. The network joining thousands of nodes carries teraoctets per second across the whole machine, though no single link does. The design problem is to arrange the topology so that any node can reach any other quickly, which is why these networks are built as multi-dimensional meshes and trees rather than as simple stars.
To make the number concrete, one teraoctet per second would fill a large consumer hard drive in twenty seconds and transfer the entire text of every book in a national library within a minute. Nothing in ordinary use approaches it, and no external cable of any kind carries it.
The unit also appears in descriptions of storage arrays. A large parallel filesystem, spread across thousands of drives, can deliver a few teraoctets per second in aggregate to a supercomputer, and that figure is quoted as a headline specification because it determines how quickly a simulation can be written out and read back.
One teraoctet per second equals 1,000 gigaoctets per second, 8 terabits per second, or about 0.9095 tebioctets per second.