How long a file takes to download given size, bandwidth and network efficiency.
Time = file_bits / (bandwidth_bps × efficiency). 1 byte = 8 bits. 1 Mbps = 1 000 000 bps (decimal); 1 MB = 1 048 576 bytes (binary). Efficiency accounts for TCP overhead, latency and shared-network contention — typical 70–85 % on broadband, 50–70 % on mobile.
Download (or upload) time is the unit-aware question every user asks at least weekly: "Will this 5 GB game download on my hotel Wi-Fi before checkout?" The math is trivial in the abstract — file size divided by bandwidth — but is plagued by unit confusion (MB vs Mbps, decimal kilo vs binary kilobinary), real-world overhead (TCP, packet loss, server-side throttling), and shared-network contention. A naive estimate using nameplate bandwidth is consistently off by 20–40 %; a calculator that exposes a fudge-factor (network efficiency) and converts units cleanly gives a usable real-world answer instead of a marketing-claim answer.
Time (seconds) = file_size_bits / (bandwidth_bps × efficiency).
Units to canonicalize:
Efficiency (0–100 %) is a multiplier that accounts for: - TCP overhead (headers, ACKs, congestion control): ~3–5 %. - Packet loss and retransmits: 0–10 % depending on link quality. - Server-side throttling: 0–50 % depending on platform. - Shared-network contention (your housemates streaming Netflix): variable.
Typical real-world efficiency: 70–85 % on broadband fiber, 50–70 % on 4G mobile, 30–60 % on shared coffee-shop Wi-Fi.
The calc converts the result to HH:MM:SS form for human readability, plus the raw seconds number, the effective throughput in MB/s, the file size in MB for sanity, and a timeline visualization with 25/50/75 % ticks.
Enter the file size with its unit (B, KB, MB, GB, TB). Enter the bandwidth with its unit (Kbps, Mbps, Gbps for the bit-based ISP plan numbers; MB/s or GB/s for byte-based throughput numbers like SSD speed tests). Set the network efficiency (default 80 %; tune lower for mobile or shared connections). The headline result is the formatted duration; the timeline visualization gives a quick visual sense of the wait.
1 GB file on a 500 Mbps fiber connection at 80 % efficiency.
Movie download (4 GB) on 4G mobile at 30 Mbps × 70 %.
5 MB photo on 5G at 200 Mbps × 75 %.
MB vs Mb / B vs b. Capital B = byte; lowercase b = bit. 1 byte = 8 bits. A "100 Mbps" connection delivers max 12.5 MB/s, not 100 MB/s. The calc handles both; users routinely confuse them.
Decimal vs binary. ISPs use decimal mega (1 Mbps = 1 000 000 bps) for marketing. Storage uses binary mega (1 MB = 1 048 576 B). A "50 Mbps for 50 MB file" is not 1 second — it's 8 seconds (size × 8 bits/byte / bandwidth = 50 × 1 048 576 × 8 / 50 000 000 ≈ 8.39 s).
ISP throttling. After a "fair use" data cap (often 100–500 GB/month on cable plans), bandwidth can be throttled to 1 Mbps or less. The nameplate speed is the un-throttled max.
Server-side limits. A 1 Gbps connection downloading from a server that caps users at 50 MB/s is bandwidth-limited by the server, not your link. The calc's efficiency parameter approximates this.
Latency vs bandwidth. For small files, latency (round-trip-time, 10–100 ms) dominates over bandwidth: 1 KB transfer over a 1 Gbps link with 50 ms latency takes ~50 ms total, not the 8 µs the bandwidth alone would suggest. The calc assumes large enough files for bandwidth to dominate (typical for media, software, datasets).
Burst vs sustained. Wi-Fi adapters report peak instantaneous throughput; sustained transfer is lower. Use the "effective throughput" output, not the connection nameplate.
TCP slow-start. New TCP connections start with a small window and double it each RTT until reaching the link capacity. Total time for small files is dominated by ramp-up, not steady-state bandwidth.
Multipath / multiple connections. Browsers open 6+ parallel connections per host; some clients (BitTorrent, parallelized downloaders) saturate the link via concurrency. The calc treats single-stream throughput.
Shared media (Wi-Fi, cable). Wi-Fi is half-duplex and shared among connected devices; cable internet is shared among neighborhood subscribers in some plant types. Effective bandwidth drops at peak hours.
HTTP overhead. HTTPS adds 1–2 KB of handshake; HTTP/2 multiplexes; HTTP/3 (QUIC) avoids TCP-handshake. For small files (< 100 KB), protocol overhead matters; for large files, negligible.
Resume vs restart. If the download fails partway, modern HTTP supports range requests for resume. The calc assumes a clean uninterrupted run.