Kibibit to Tebibyte

A kilobit (kb) = 1,000 bits (SI decimal). A kibibit (Kibit) = 1,024 bits (IEC binary). The difference is 24 bits (2.4%) — small but matters in precise hardware specifications. The kibibit was introduced in 1998 to provide an unambiguous binary unit, since networking engineers had been using "kilobit" to mean both 1,000 and 1,024 bits in different contexts.

For decades, computer engineers used SI prefixes (kilo-, mega-, giga-) to mean powers of 1,024 in binary contexts and powers of 1,000 in SI/metric contexts. This caused real confusion: a "64 kilobyte" RAM chip had 65,536 bytes, while a "64 kilobyte" internet packet had 64,000 bytes. The IEC defined kibi- (1,024), mebi- (1,048,576), etc. in 1998 to give engineers unambiguous binary units.

Kibibits are rarely used directly in OS user interfaces — OSes work in bytes and their binary multiples (KiB, MiB, GiB). Kibibits appear in hardware documentation, FPGA bitstream sizes, and some network protocol headers where binary bit counts matter. Network speeds remain in decimal kilobits per second even in technical contexts.

Before IEC 80000-13 (1998), "kilobit" meant either 1,000 or 1,024 bits depending on context — RAM datasheets used 1,024 while telecom specs used 1,000. The IEC standard introduced kibibit (1,024 bits) as the unambiguous binary term, reserving kilobit strictly for 1,000 bits. Adoption took over a decade: Linux adopted IEC prefixes around 2010, and JEDEC still allows the old dual-meaning convention for memory marketing.

IEC binary prefixes have been slowly adopted: Linux tools (df, free) now use GiB and MiB; macOS used decimal GB since 2009; Windows switched to GiB labeling in Windows 10/11. However, kibibit specifically remains a niche technical term — consumer-facing software almost never uses it. Engineers working on embedded systems, FPGAs, and memory hardware are its primary audience.

TB (terabyte) = 10¹² bytes = 1,000,000,000,000 bytes (SI decimal). TiB (tebibyte) = 2⁴⁰ bytes = 1,099,511,627,776 bytes (IEC binary). TiB is 9.95% larger. The practical consequence: a 1 TB hard drive (decimal) holds 0.9095 TiB. This 10% gap is the primary reason drive capacity appears lower in the OS than on the box.

ZFS and Btrfs are copy-on-write filesystems designed for TiB-scale pools with built-in features that traditional filesystems lack. ZFS supports inline deduplication — a 10 TiB pool with 40% duplicate data might show 6 TiB of logical usage but only consume 3.6 TiB physically. Btrfs offers transparent compression (zstd), where a 4 TiB dataset of compressible log files might occupy only 1–2 TiB on disk. Both support snapshots that initially consume zero extra space, growing only as data diverges. These features make "used space in TiB" surprisingly complex to report accurately.

Yes. Linux tools (df -h, lsblk) display storage in IEC binary units: KiB, MiB, GiB, TiB. df -h output showing "1.8T" for a 2 TB drive is reporting 1.8 TiB. Modern Linux distributions correctly label these as TiB in technical contexts. This is one of the areas where Linux is more technically precise than Windows or consumer storage labels.

RAID arrays lose capacity to redundancy: RAID 1 mirrors two drives (50% efficiency); RAID 5 loses one drive worth of capacity; RAID 6 loses two drives. A 4-drive RAID 5 array of 2 TB drives has 3 × 2 TB = 6 TB raw usable (decimal), ≈ 5.46 TiB, minus filesystem overhead. Enterprise storage also reserves space for spares, snapshots, and wear levelling, further reducing usable TiB.

No. A tebibyte (TiB) = 2⁴⁰ bytes = 1,099,511,627,776 bytes — about 1.1 trillion bytes. Exactly one trillion bytes = 10¹² bytes = 1 terabyte (TB, decimal). The tebibyte is approximately 10% larger than a trillion bytes. "Terabyte" is often casually used to mean "1 trillion bytes"; "tebibyte" is the precise binary equivalent at 1,024 gibibytes.