How this calculation works
Normalize payload to bits and rate to bits per second. Apply your utilization assumption once, divide the total batch payload by effective throughput, then add one shared startup allowance. A rate comparison keeps payload and startup unchanged.
Total seconds = file size × bytes/unit × file count × 8 / (bit rate × utilization/100) + batch startup secondsA worked example
One 1 GB file contains 8 billion bits. At 100 Mbps with an entered 80% payload utilization, effective throughput is 80 million bit/s and payload time is 100 seconds. Adding a single 5-second startup gives 105 seconds.
Good to know
An entered-rate estimate, not a network test. Files are equal sized and share one total rate. No measurement, download, compression, congestion, retries, queue or per-file startup simulation. Decimal and binary units are separate; numeric results are rounded for display.
Reference: NIST: decimal and binary prefixes, bits and bytes (opens in a new tab) ↗
One capital letter can change the estimate eightfold
A byte contains eight bits. A nominal 100 Mbps is 100 million bits per second, whereas 100 MB/s is 100 million bytes per second. Before adding utilization or startup assumptions, that is an eightfold difference in the payload rate. The selector spells out each definition so the letter case is not left to guesswork.
Decimal prefixes and binary prefixes answer another unit question. One GB is 1,000,000,000 bytes; one GiB is 1,073,741,824 bytes. With the default 100 Mbps rate, 80% utilization and five-second startup, a 1 GB payload takes 105 seconds under the model. Replacing GB with GiB gives approximately 112.374 seconds. The network did not change; the entered payload did.
Use the digital storage converter to compare file-size units separately. This calculator adds an explicit throughput and time model. It never downloads a file of the entered size, so estimating the transfer of a large backup does not consume that backup’s worth of data.
The digital storage converter compares payload units. Data transfer time also needs a bit rate, an explicit utilization assumption and any separately entered startup allowance.
Read more: NIST: distinguish decimal prefixes, binary prefixes and bit/byte units (opens in a new tab) ↗
Choose a utilization assumption that matches the rate you entered
The default 80% is an editable example, not a measured overhead or a promise about any connection. Effective payload rate is nominal rate multiplied by that fraction. For 100 Mbps at 80%, the model uses 80 Mbps. It does not add another hidden reduction after that multiplication.
A measured application transfer rate may already reflect overhead, congestion and server limits. Applying 80% to that measurement would reduce it a second time. Use 100% when the input already describes the payload rate you intend to assume, unless you deliberately want an additional margin. A speed-test result and an application download need not sustain the same rate.
For equal-sized files, multiply the payload by the file count and use the total shared rate once. Ten files do not each receive the full connection rate in this model. Sending them concurrently can change real scheduling, but it does not create ten times the aggregate bandwidth just because ten file transfers are active.
A fixed startup changes how much faster a faster link feels
At the default inputs the payload needs 100 seconds and the single batch allowance adds five seconds. Doubling the entered rate halves payload time to 50 seconds, giving 55 seconds total. The total is not 52.5 seconds because the five-second startup does not halve with the rate.
The comparison table keeps the same payload, utilization and batch allowance while trying half, equal and double the nominal rate. Its rows are scenarios, not repeat speed-test measurements. For a very small file, an entered startup allowance can dominate. A zero-size payload deliberately leaves that allowance in place rather than reporting a transfer measurement.
The main result retains fractional seconds. The readable days/hours/minutes/seconds display rounds the duration up to the next whole second, which avoids showing zero seconds for a positive sub-second estimate. The calculation assumes a constant rate throughout; pauses, retries, compression, queues and per-file setup need a different or more detailed model.
A few questions, answered
Does this run an Internet speed test?
No. It calculates an estimate from your entered file sizes, rate and assumptions. It does not transfer the modeled payload or measure the network.
Are Mbps and MB/s interchangeable?
No. Mbps describes megabits per second; MB/s describes megabytes per second. One byte contains eight bits, so the same numerical MB/s value is eight times the bit rate.
Is the startup allowance charged for every file?
No. It is added once for the whole batch. All files share the entered aggregate rate and have the same entered size.
Why can the duration differ from an actual download?
The model uses a constant assumed payload rate. Real throughput, server behavior, setup, queues, retries and compression can vary. The utilization percentage is an input, not a measurement.