| Megabits per second (Mbit/s) | Teraoctets per second (To/s) |
|---|---|
| 1 Megabit per second | 0.000000125 To/s |
| 2 Megabits per second | 0.00000025 To/s |
| 3 Megabits per second | 0.000000375 To/s |
| 4 Megabits per second | 0.0000005 To/s |
| 5 Megabits per second | 0.000000625 To/s |
| 10 Megabits per second | 0.00000125 To/s |
| 20 Megabits per second | 0.0000025 To/s |
| 25 Megabits per second | 0.000003125 To/s |
| 50 Megabits per second | 0.00000625 To/s |
| 100 Megabits per second | 0.0000125 To/s |
| Reference | Megabits per second (Mbit/s) | Teraoctets per second (To/s) |
|---|---|---|
| A dial-up modem | 0.056 Mbit/s | 0.000000007 To/s |
| Typical home broadband | 100 Mbit/s | 0.0000125 To/s |
| Gigabit Ethernet | 1000 Mbit/s | 0.000125 To/s |
| Streaming a 4K film | 25 Mbit/s | 0.000003125 To/s |
The megabit per second is a unit of data transfer rate equal to one million bits per second. Its symbol is Mbit/s, often written Mbps. It is the unit in which internet connections are sold, which makes it the data unit most people encounter by name.
Because it is a decimal million and not 1,048,576, the conversion to octets is exact and easy: one megabit per second is 125 kilooctets per second, so a hundred-megabit connection delivers about 12.5 megaoctets per second at best. Anyone who watches a file transfer and does the division has understood the entire relationship between how connections are advertised and how transfers are reported.
What a household actually needs is far below what it usually buys. Standard-definition video streaming uses about 3 megabits per second, high definition about 5, and ultra-high definition about 25. A video call is around 3 to 8. A large family watching four separate high-definition streams while somebody downloads a game is using perhaps 60 megabits per second, which a hundred-megabit connection handles comfortably.
The reason to buy more capacity than that is not peak speed but behaviour under load. A link that is near its limit develops queues, and queues add delay, which shows up as stutter in video calls and lag in games. A connection with generous headroom keeps its latency low, and that is a more noticeable improvement than a higher number on a speed test.
Wired local networks pass through this range on the way up. The original Ethernet ran at 10 megabits per second, its successor at 100, and both were the standard office connection for a decade each before gigabit replaced them. Wireless standards followed the same path with a lag, and both are now measured in hundreds of megabits or in gigabits.
Real throughput is always below the nominal rate. Protocol overhead takes 5 to 10 per cent on a wired link; a shared wireless channel loses much more, because the medium is divided between all the devices using it and interference forces retransmission. A connection advertised at 100 megabits per second measured at 90 over cable and 50 over a busy wireless network is behaving normally.
One megabit per second equals 1,000,000 bits per second, 125 kilooctets per second, or about 0.9537 mebibits 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.