Binary prefixes are unit prefixes based on powers of 1,024 rather than 1,000: kibi (Ki), mebi (Mi), gibi (Gi), tebi (Ti), pebi (Pi) and onward. A kibibyte (KiB) is 1,024 bytes, a mebibyte (MiB) is 1,024 KiB and a gibibyte (GiB) is 1,024 MiB. The International Electrotechnical Commission (IEC) introduced them in the late 1990s so that binary quantities would stop borrowing the decimal prefixes kilo, mega and giga, which in the International System of Units (SI) mean exactly 1,000, 1,000,000 and 1,000,000,000. The mismatch between the two conventions is why storage capacity, operating system readouts and cloud bills often appear not to agree.
At a glance
- Decimal (SI) prefixes step by 1,000: a kilobyte is 1,000 bytes and a terabyte is 1,000,000,000,000 bytes.
- Binary (IEC) prefixes step by 1,024: a KiB is 1,024 bytes, a MiB is 1,048,576 bytes and a GiB is 1,073,741,824 bytes.
- The gap grows with size: about 2.4% at kilo, 4.9% at mega, 7.4% at giga and 10% at tera.
- A drive sold as 1 TB holds about 931 GiB, which Windows displays as “931 GB” because it labels binary amounts with decimal prefixes.
- Networking uses decimal prefixes almost without exception, while memory sizes use binary amounts by convention.
What problem it solves
Computers address memory in powers of two, so early engineers found it convenient to call 1,024 bytes a “kilobyte” because 1,024 is close to 1,000. As capacities grew, the gap between the binary meaning and the decimal meaning grew with them, and the same label came to mean two different amounts. A gigabyte might be 1,000,000,000 bytes or 1,073,741,824 bytes depending on who wrote it.
Binary prefixes give the binary amounts their own names, so “GiB” always means 1,024 cubed bytes and “GB” means 1,000 cubed. Adoption is incomplete, which is the practical problem for buyers: the same storage, transfer or capacity figure can be presented both ways, and the labels do not always tell you which.
How it works
The two series.
| Decimal (SI) | Value | Binary (IEC) | Value |
|---|---|---|---|
| kilobyte (kB) | 1,000 bytes | kibibyte (KiB) | 1,024 bytes |
| megabyte (MB) | 1,000² bytes | mebibyte (MiB) | 1,024² bytes |
| gigabyte (GB) | 1,000³ bytes | gibibyte (GiB) | 1,024³ bytes |
| terabyte (TB) | 1,000⁴ bytes | tebibyte (TiB) | 1,024⁴ bytes |
| petabyte (PB) | 1,000⁵ bytes | pebibyte (PiB) | 1,024⁵ bytes |
The series continues through exabyte, zettabyte and yottabyte on the decimal side, with matching binary names (exbi, zebi, yobi).
The 1 TB drive example. A drive sold as 1 TB contains 1,000,000,000,000 bytes. Divide by 1,024 four times and you get about 0.91 TiB, or about 931 GiB. Windows reports that as roughly 931 GB. Recent versions of macOS use decimal units and report close to 1 TB. Linux tools vary: some show binary units with Ki/Mi/Gi labels and others can show either. The bytes are the same in every case.
Other reasons usable space is lower. Separately from the units, the file system takes space for its own structures, some file systems reserve a share for administrators, and RAID or erasure coding in storage systems consumes capacity for redundancy. Vendors often quote raw capacity, so usable capacity can be much lower than the headline figure.
Where each convention shows up. Drive and storage marketing uses decimal units. RAM is sized in binary amounts. Network speeds and bits per second are decimal. Cloud providers and storage platforms vary by product, with some stating GiB and others GB in their documentation.
When it matters for buyers
- Sizing storage purchases. When comparing NAS, SAN or block storage quotes, convert every figure to the same unit and to usable capacity. A 10% unit difference at the terabyte scale can be the gap between enough and not enough. Our file and object storage page covers how buyers compare storage services.
- Reading cloud bills. Storage, snapshots and data egress fees are billed per GB or GiB. Check which the price list means before comparing providers.
- Matching storage to bandwidth. Transfer allowances are usually in bytes and circuits in bits per second, often with different prefix conventions on each side. See byte for converting between them.
- Migrations and backups. Estimating how long a copy will take or how much target capacity you need goes wrong if the source and target report sizes in different units.
- Comparing file and object platforms. File systems and object storage services may report usage differently from your on-premises tools.
Questions to ask vendors
- Are your capacity figures decimal (GB, TB) or binary (GiB, TiB)?
- Is the quoted capacity raw or usable after RAID, erasure coding, formatting and reserved space?
- How is a GB defined on your price list for storage and data transfer?
- Which units do your management console and usage reports display, and do they match the billing units?
- If we exceed a capacity commitment, is the overage measured in the same unit as the commitment?
How it differs from decimal (SI) prefixes
Decimal prefixes, defined by the International System of Units, always step by 1,000: kilo is 1,000, mega is 1,000,000 and giga is 1,000,000,000. Binary prefixes step by 1,024 and use different names and symbols, such as kibi (Ki) and gibi (Gi). Strictly, “GB” means 1,000³ bytes and “GiB” means 1,024³ bytes, but much software still writes GB for binary amounts, so the label alone does not always settle which one is meant.