| Teraoctets per second (To/s) | Megaoctets per second (Mo/s) |
|---|---|
| 1 Teraoctet per second | 1000000 Mo/s |
| 2 Teraoctets per second | 2000000 Mo/s |
| 3 Teraoctets per second | 3000000 Mo/s |
| 4 Teraoctets per second | 4000000 Mo/s |
| 5 Teraoctets per second | 5000000 Mo/s |
| 10 Teraoctets per second | 10000000 Mo/s |
| 20 Teraoctets per second | 20000000 Mo/s |
| 25 Teraoctets per second | 25000000 Mo/s |
| 50 Teraoctets per second | 50000000 Mo/s |
| 100 Teraoctets per second | 100000000 Mo/s |
| Reference | Teraoctets per second (To/s) | Megaoctets per second (Mo/s) |
|---|---|---|
| A dial-up modem | 0.000000007 To/s | 0.007 Mo/s |
| Typical home broadband | 0.0000125 To/s | 12.5 Mo/s |
| Gigabit Ethernet | 0.000125 To/s | 125 Mo/s |
| Streaming a 4K film | 0.000003125 To/s | 3.125 Mo/s |
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.
The megaoctet per second is a unit of data transfer rate equal to one million octets per second, or eight megabits per second. Its symbol is Mo/s. It is the unit in which storage devices and their interfaces are rated, and the one a file manager shows while a copy is running.
Storage speeds fall naturally into this range. A mechanical hard drive sustains 100 to 250 megaoctets per second on sequential reads. A solid-state drive on the older interface reaches about 550, which is the limit of that interface rather than of the drive. A modern drive on the faster interface reaches several thousand, at which point the unit gives way to gigaoctets per second.
Interface speeds tell the same story from the other side. The successive versions of the universal serial bus have offered roughly 1.5, 60, 625 and 2,500 megaoctets per second at their nominal rates, and memory card standards have followed a similar path. A device is always limited by whichever of the two is slower, and matching them is the practical art of building a fast system.
For everyday sizes, one megaoctet per second transfers a photograph in a couple of seconds and a two-gigaoctet film in about half an hour. At 500 megaoctets per second the same film takes four seconds. That contrast explains why the perceived speed of a computer changed so completely when solid-state storage replaced mechanical drives, even though processors improved far less over the same period.
Sequential rates like these are the best case. Reading many small files instead of one large one costs far more, because each file requires locating its data and reading its record. A drive that sustains 500 megaoctets per second in sequence may manage only a few tens when copying a directory of thousands of small files, and that is where the difference between drive technologies is most visible.
The unit also describes the throughput of a busy network connection: a gigabit link delivers 125 megaoctets per second at best, so a fast local network and a fast drive are now comparable, and neither is obviously the limiting factor in a transfer.
One megaoctet per second equals 1,000,000 octets per second, 8 megabits per second, or about 0.9537 mebioctets per second.