Gigabyte to Pebibyte

Hard drive manufacturers measure 1 TB as 1,000,000,000,000 bytes (decimal). Windows displays storage in gibibytes (binary) but historically labelled them as "GB" — so 1,000,000,000,000 bytes ÷ 1,073,741,824 ≈ 931 GiB, which Windows displayed as "931 GB". macOS (since 10.6) correctly reports the same drive as "1 TB" using decimal GB. The drive is not lying; the OS was using a binary unit with a decimal label.

8 GB RAM is the current minimum for gaming; 16 GB is the recommended standard for most modern games at 1080p and 1440p; 32 GB benefits heavily multitasking systems or games with large open worlds. Memory-intensive tasks like video editing, 3D rendering, and running large language models locally typically require 32–64 GB or more.

A 4K movie in H.264 or H.265 encoding is typically 50–100 GB on Blu-ray; streaming services compress aggressively to 15–25 GB for 4K HDR content. Netflix's 4K streams average about 7 GB per hour; the downloaded version via the Netflix app for offline viewing is roughly 3–6 GB per hour at high quality settings.

1 GB of mobile data supports roughly: 2–3 hours of music streaming, 1 hour of HD video streaming, 2–3 hours of web browsing, or 30–60 minutes of video calling. Social media apps with autoplay video are heavy consumers — TikTok and Instagram Reels can use 300–600 MB per hour of active use.

AI model sizes vary enormously. GPT-2 (2019) is about 1.5 GB; Llama 2 7B is roughly 13 GB in float16 precision; Llama 2 70B is about 130 GB. GPT-4-class models are estimated at 500+ GB. Quantised (compressed) versions are smaller: a 4-bit quantised 7B model fits in about 4 GB, runnable on a modern laptop. Training requires far more — the training dataset, gradients, and optimizer states for a 70B model can occupy 1–2 TB of GPU memory across a cluster. The trend toward larger models is driving consumer GPU memory from 8 GB to 16–24 GB as a baseline for local AI inference.

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.