Convert GB to TB

A gigabyte is a billion bytes in the decimal sense used by drive manufacturers, phone plans and video sizes.

A terabyte is a trillion bytes. A drive sold as 1 TB is exactly that — the space that seems to go missing is a unit disagreement, not a defect.

Going from gigabytes to terabytes moves the decimal point 3 places to the left and changes nothing else. There is no factor to remember and no rounding to decide: the digits stay in the same order and only their position changes.

1,234 GB is 1.234 TB — same digits, moved along. That is worth knowing because it is the one kind of conversion you can check at a glance: if the digits of the answer are not the digits you started with, something other than the conversion has happened to them.

Storage capacity is decimal and always has been. Tera is the SI prefix for 10¹², so a drive sold as 1 TB contains 1,000,000,000,000 bytes and one sold as 4 TB contains four times that. Adding gigabyte figures up and dividing by a thousand therefore gives the number to shop with directly: 2,400 GB of material needs a 4 TB drive rather than a 2 TB one, because 2.4 TB does not exist as a product.

The binary system is not what capacity is sold in, and on this page it matters only as a translation at the end. A tebibyte is 1,024 gibibytes, which is 1,099,511,627,776 bytes and 10 per cent more than a terabyte. That is the widest gap among the prefixes most people meet, and each step above it is wider still — pebi against peta is 12.6 per cent — which is why a purchase sized in the wrong system can be a whole tier short.

The step from a pile of drives to usable capacity is subtraction, and it is not a percentage. Four 1 TB drives striped present nearly 4 TB; the same four with single parity present about 3 TB, because one drive’s worth of capacity is spent on the ability to lose a drive; with double parity it is 2 TB; mirrored, it is exactly half. The cost of redundancy is fixed per array rather than proportional, which is why it is brutal on small arrays and mild on large ones.

That changes how a purchase should be sized. Needing 3 TB of usable space with single parity means buying four 1 TB drives or three 2 TB drives, and the two options differ in cost, in rebuild time and in how much is lost when a second drive fails during the rebuild. Do the gigabyte arithmetic on the usable figure first, then work backwards to the drives, rather than adding up labels and discovering the parity cost afterwards.

A freshly formatted volume already has less free space than the drive has capacity. The filesystem writes its own structures — allocation tables, inode tables, journals — and on a large volume that is a few gigabytes. Several filesystems also reserve a fraction of capacity for privileged processes by default, historically five per cent, which on a 4 TB volume is 200 GB that never appears as available.

Solid-state drives lose a further slice before the operating system sees anything. The controller keeps spare blocks aside for wear levelling and to replace failed ones, and that over-provisioned area is deducted from the raw flash rather than added to the label. None of this is a defect and none of it is recoverable by reformatting; it is the difference between the capacity that was manufactured and the capacity that was sold to you.

The last stretch of a volume is worth less than the rest. Allocation gets harder as free space fragments, copy-on-write filesystems need room to write a new version before releasing the old one, and solid-state drives need free blocks to write into without an erase cycle first. The practical effect is that write performance falls off well before a volume reports itself full, and on some filesystems the drop is steep rather than gradual.

So the sizing rule is to plan on filling about 80 per cent of what you buy. A requirement of 3,200 GB is not a 4 TB purchase with 800 GB to spare; it is a 4 TB purchase working at the top of its comfortable range with no room for growth. Expressed the other way, dividing the gigabyte requirement by 0.8 before converting to terabytes gives a figure that will still be usable in two years.

A capacity decision is a forecast, and the gigabyte figure people start from is usually the current one. Add the growth rate over the intended life of the purchase before converting: a collection at 1,800 GB growing 25 GB a month over a four-year replacement cycle reaches 3,000 GB, and that is the number to divide by a thousand rather than the 1.8 TB it is today.

Growth is also worth splitting by kind, because the rates are wildly different. Documents, mail and code grow slowly and compress well; photographs grow with every device upgrade, since sensor resolution rises; video is the term that dominates any archive it appears in, at roughly 3 GB an hour for HD and up to 7 for 4K. An archive that is mostly video is sized by hours of footage per year, and everything else is a rounding error next to it.

The two units divide by product rather than by size. Drives, arrays, NAS enclosures and cloud storage tiers are quoted in terabytes; phones, tablets, consoles, memory cards and laptop configurations are quoted in gigabytes even when the figure passes a thousand. That is why a 1 TB phone and a 1,000 GB drive are described in different units despite being the same capacity, and why a comparison across the two needs the conversion made explicit.

Cloud pricing sits in gigabytes for a different reason: it is billed per gigabyte-month, so the unit is the billing granularity rather than a product size. A tier described as 5 TB is 5,000 gigabyte-months when the invoice arrives, and comparing it against a drive means comparing a monthly rent against a one-off purchase — which is a longer calculation, but at least both sides are counting decimal gigabytes.

Once the terabyte figure is chosen, it is worth working out in advance what the system will say, so that the number is not a surprise later. Divide the byte total by 1,099,511,627,776 for a tebibyte figure: 4 TB of purchased capacity is 3.64 TiB, 8 TB is 7.28 TiB and 12 TB is 10.91 TiB. Any system counting in binary units will report figures near those, and it will very likely label them TB.

Doing that translation at purchase time turns a later shock into a confirmation. A 4 TB drive that mounts and reports 3.6 of something is behaving exactly as expected; a 4 TB drive reporting 3.2 has lost something else — a hidden recovery partition, a reserved fraction, or a formatting choice worth investigating. Knowing which number to expect is what makes the difference between the two visible.