| Tebioctets per second (Tio/s) | Kibioctets per second (Kio/s) |
|---|---|
| 1 Tebioctet per second | 1073741824 Kio/s |
| 2 Tebioctets per second | 2147483648 Kio/s |
| 3 Tebioctets per second | 3221225472 Kio/s |
| 4 Tebioctets per second | 4294967296 Kio/s |
| 5 Tebioctets per second | 5368709120 Kio/s |
| 10 Tebioctets per second | 10737418240 Kio/s |
| 20 Tebioctets per second | 21474836480 Kio/s |
| 25 Tebioctets per second | 26843545600 Kio/s |
| 50 Tebioctets per second | 53687091200 Kio/s |
| 100 Tebioctets per second | 107374182400 Kio/s |
| Reference | Tebioctets per second (Tio/s) | Kibioctets per second (Kio/s) |
|---|---|---|
| A dial-up modem | 0.00000000636646 Tio/s | 6.83594 Kio/s |
| Typical home broadband | 0.0000113687 Tio/s | 12207 Kio/s |
| Gigabit Ethernet | 0.000113687 Tio/s | 122070 Kio/s |
| Streaming a 4K film | 0.00000284217 Tio/s | 3051.76 Kio/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 kibioctet per second is a unit of data transfer rate equal to 1,024 octets per second, and therefore to 8,192 bits per second. Its symbol is Kio/s. Unlike most of the binary rate units it is genuinely common, because the command-line tools that copy, download and synchronise files have reported in it for decades.
The reason is straightforward. Those tools count what they have moved in blocks, and blocks are powers of two. A program that reads in four-kibioctet pieces and divides the total by elapsed time produces a rate in kibioctets per second, and reporting it in decimal kilooctets would require an extra multiplication for no benefit. The unit is what the arithmetic naturally produces.
Anyone who has watched a file copy on a Unix-like system has seen the figure. Download utilities, archive tools, disc-writing commands and network file transfer programs all report progress in kibioctets or mebioctets per second, and most of them label it correctly with the lowercase i. It is one of the few places where the IEC prefixes are used consistently in everyday software.
For scale, a kibioctet per second moves about a thousand characters of text each second: a short letter in a second, a novel in about ten minutes. It is a rate at which a modern web page will not load in any reasonable time, so seeing it in a progress display usually means something has gone wrong with the connection rather than that the transfer is nearly finished.
The difference from a kilooctet per second is 2.4 per cent, which nobody notices. The value of using the binary unit here is not accuracy but honesty: the number came from a binary computation, and writing it with a binary prefix says so. A reader who wants the decimal figure can convert; a reader given a decimal label for a binary number cannot recover anything.
Comparing the reading with an advertised connection speed requires two steps: multiply by eight to get bits, and adjust by 2.4 per cent for the base. In practice the second step is beneath the noise of any real measurement, and the first is the one that matters.
One kibioctet per second equals 1,024 octets per second, 8,192 bits per second, or 1.024 kilooctets per second.