| Tebioctets per second (Tio/s) | Exbibits per second (Eibit/s) |
|---|---|
| 1 Tebioctet per second | 0.00000762939453125 Eibit/s |
| 2 Tebioctets per second | 0.0000152587890625 Eibit/s |
| 3 Tebioctets per second | 0.0000228881835938 Eibit/s |
| 4 Tebioctets per second | 0.000030517578125 Eibit/s |
| 5 Tebioctets per second | 0.0000381469726562 Eibit/s |
| 10 Tebioctets per second | 0.0000762939453125 Eibit/s |
| 20 Tebioctets per second | 0.000152587890625 Eibit/s |
| 25 Tebioctets per second | 0.000190734863281 Eibit/s |
| 50 Tebioctets per second | 0.000381469726562 Eibit/s |
| 100 Tebioctets per second | 0.000762939453125 Eibit/s |
| Reference | Tebioctets per second (Tio/s) | Exbibits per second (Eibit/s) |
|---|---|---|
| A dial-up modem | 0.00000000636646 Tio/s | 4.85723 × 10-14 Eibit/s |
| Typical home broadband | 0.0000113687 Tio/s | 8.67362 × 10-11 Eibit/s |
| Gigabit Ethernet | 0.000113687 Tio/s | 8.67362 × 10-10 Eibit/s |
| Streaming a 4K film | 0.00000284217 Tio/s | 2.1684 × 10-11 Eibit/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 exbibit per second is a unit of data transfer rate equal to two to the sixtieth power bits per second, which is 1,024 pebibits per second. Its symbol is Eibit/s. It is the binary counterpart of the exabit per second, and the two differ by 15.3 per cent.
That fifteen per cent is more than the margin most engineering work allows anywhere. A quantity stated in the wrong convention at this level is not slightly imprecise but plainly wrong, and no amount of context recovers the intended figure once it has been written ambiguously. This is the situation the IEC prefixes were introduced to prevent.
No physical link approaches this rate. The unit describes aggregates: the total instantaneous traffic of the whole internet is roughly an exabit per second, so a binary exbibit per second is fifteen per cent more than everything moving on every network on the planet at a given moment.
Where a binary figure of this size arises naturally is in address arithmetic rather than in transmission. Two to the sixtieth is a number that appears throughout the design of 64-bit systems, and any rate derived from filling or traversing such an address space in a fixed time is naturally binary. Those calculations are theoretical, but they are the reason the unit is defined rather than left unnamed.
In octets an exbibit per second is 144,115,188,075,855,872, or 128 pebioctets per second. That is more storage moved per second than the world manufactures in several years, which is a way of saying that no source and no destination for such a flow could exist.
For a converter the treatment is mechanical and must be exact. Eibit/s and Ebit/s differ by an eighth, and a tool that treats them as interchangeable has introduced an error larger than the difference between many neighbouring units. The lowercase i is not decoration.
One exbibit per second equals 1,024 pebibits per second, 144,115,188,075,855,872 octets per second, or about 1.153 exabits per second.