What does this calculator do?
The bandwidth-delay product is the amount of data that fits "on the wire" between two endpoints: link rate multiplied by round-trip time. It matters because a TCP sender may only have one window of unacknowledged data outstanding. If the window is smaller than the BDP, the sender stalls waiting for acknowledgements and the link never reaches its rated speed, no matter how much bandwidth you buy. This is the classic long-fat-network problem on high-speed intercontinental paths. The result is exact arithmetic on the values you give; whether your stack can actually use a window that large depends on its configuration. Units follow the standards exactly: kB, MB, GB and TB are powers of ten (1 GB = 10⁹ bytes) while KiB, MiB, GiB and TiB are powers of two (1 GiB = 2³⁰ bytes). The two ladders are never mixed, so "1 GB = 1024 MB" is not treated as a universal rule.
Formula
Multiply the rate in bits per second by the RTT in seconds, then convert to bytes.
bits = bandwidthBitsPerSecond × (rttMilliseconds ÷ 1000) bytes = bits ÷ 8 KiB = bytes ÷ 2^10
How to use it
Enter the bandwidth of the bottleneck link with its unit, and the round-trip time in milliseconds — the figure a ping or a traceroute to the far end reports, measured when the path is not congested. The bytes-in-flight result is the minimum TCP receive window and send buffer required to saturate the path with a single stream. This calculator does no measuring of its own: the RTT is a number you supply, not something it probes.
Worked example
A 1,000 Mbps link with an 80 ms RTT: 1 × 10⁹ bit/s × 0.08 s = 80,000,000 bits in flight, which is 10,000,000 bytes — 10 MB, or about 9,765.63 KiB. A single TCP stream with a default 64 KiB window would reach only a small fraction of that link; window scaling (RFC 7323) is required.
Frequently asked questions
My link is fast but a single transfer is slow — why?
Most likely the TCP window is smaller than the BDP. The sender fills the window, then waits a full round trip for acknowledgements. Enable window scaling, raise the socket buffers, or use several parallel streams.
Should I use KiB or MB when configuring buffers?
Operating-system socket buffer settings are usually in bytes, so use the bytes figure directly. KiB is shown because tuning guides often quote window sizes in KiB, and MB is there for a quick sense of scale.
Does this account for packet loss?
No. The BDP is the pure geometry of the path. Loss makes matters worse: classic TCP congestion control halves the window on loss, so a lossy long-distance path can sit far below the BDP even with large buffers.
Sources
- NIST — Prefixes for binary multiples (IEC 80000-13): kibi/mebi/gibi are powers of 1024, while kilo/mega/giga are powers of 1000. This is why Mbit/s, MB/s and MiB/s are never mixed here.
- BIPM — The International System of Units (SI Brochure): decimal prefixes (k = 10³, M = 10⁶, G = 10⁹) and the second, the basis of every data-rate and transfer-time conversion.
- RFC 7323 — TCP Extensions for High Performance: the bandwidth-delay product and why a link needs a window of that size to stay full.
- Reviewed: 2026-08-10
Related calculators
Last updated: 2026-08-10 · v1.0.0 · Report an error