Pebibyte to Gibibit

PB (petabyte) = 10¹⁵ bytes = 1,000,000,000,000,000 bytes (SI decimal). PiB (pebibyte) = 2⁵⁰ bytes = 1,125,899,906,842,624 bytes (IEC binary). PiB is 12.59% larger. For a data center purchasing 100 PiB of raw storage, the SI vs IEC confusion would represent approximately 12.59 PB of missing or unexpected capacity.

Cloud providers (AWS, Azure, GCP) operate at exabyte scale but provision and bill individual customers at PiB scale for enterprise storage. Scientific computing facilities like CERN, the Square Kilometer Array telescope project, and US national laboratories store tens to hundreds of PiB. Large video platforms (Netflix, YouTube) store hundreds of PiB of encoded video content.

Using 20 TB drives (a 2024 high-density consumer drive): 1 PiB = 1,125,899,906,842,624 bytes ÷ 20,000,000,000,000 bytes/drive ≈ 56.3 drives. So roughly 57 × 20 TB drives to fill 1 PiB. In a data center using 60-drive storage shelves, one shelf of 60 × 20 TB drives provides about 1.07 PiB of raw capacity.

Magnetic tape (LTO technology) remains the dominant medium for cold storage at PiB scale due to economics and durability. An LTO-9 cartridge holds 18 TB (uncompressed) and costs roughly $100 — about $5.50 per TB, versus $15–20 per TB for HDDs. Tape also consumes zero power when idle, unlike spinning disks. The IBM TS4500 tape library can hold over 40 PiB in a single rack. Major users include CERN, national archives, and film studios — Netflix stores its master copies on tape. Tape's main downside is sequential access: retrieving a specific file can take minutes versus milliseconds for disk.

CERN's Worldwide LHC Computing Grid stores approximately 300–400 PB (petabytes, decimal) of data across distributed sites, with the main Tier-0 facility at CERN holding about 100 PB on disk and 200 PB on tape. The LHC generates roughly 15 PB of data per year from collision events. Future upgrades (High-Luminosity LHC) are projected to increase this to 50–100 PB per year.

A gigabit (Gbit) = 10⁹ bits = 1,000,000,000 bits (SI). A gibibit (Gibit) = 2³⁰ bits = 1,073,741,824 bits (IEC binary). The difference is 7.37%. Consumer networking equipment and ISP speed ratings use decimal gigabits; memory and chip designers sometimes use gibibits when binary precision is required.

Virtually all networking equipment — routers, switches, NICs, ISP speed ratings — uses decimal gigabits (Gbit). A "1 Gbps" (gigabit per second) connection means exactly 1,000,000,000 bits per second, not 1,073,741,824 bits per second. Network standards (Ethernet IEEE 802.3) are defined in SI units.

DDR memory bandwidth is calculated from clock speed, bus width, and transfers per clock. A DDR5-4800 module on a 64-bit bus delivers 4,800 MT/s × 64 bits = 307,200 Mbit/s ≈ 292.97 Gibit/s. Engineers use gibibits when verifying that memory throughput matches binary-aligned cache line sizes (typically 512 bits = 64 bytes), ensuring no fractional transfers occur during burst reads.

GPU memory bandwidth is typically quoted in gigabytes per second (GB/s) using SI decimal values — not gibibits. For example, NVIDIA's RTX 4090 has 1,008 GB/s of memory bandwidth (decimal). Some academic papers and IEEE publications convert this to GiB/s or Gibit/s for precision, but consumer GPU marketing universally uses SI decimal units.

Gibibit appears in: IEEE standards documents specifying memory interface speeds, JEDEC memory specifications, some academic networking papers, and storage controller datasheets. Consumer-facing software, marketing materials, and OS interfaces virtually never display gibibits — they show gigabits (networking) or gigabytes (storage). It is primarily a precision engineering unit.