Bits and bytes

Network plans commonly advertise megabits per second, written Mbps with a lowercase b. File sizes commonly use bytes, with uppercase B. One byte contains eight bits, so a 100 Mbps line has a theoretical maximum of 12.5 decimal MB/s before overhead. Confusing the letters produces an eightfold time error.

The calculator converts the file to bytes, multiplies by eight and divides by bits per second. It then divides by the throughput efficiency to model a connection that does not deliver line rate for the whole transfer.

Decimal and binary file units

Decimal MB and GB use powers of 1000: one GB is one billion bytes. Binary MiB and GiB use powers of 1024: one GiB is 1,073,741,824 bytes. Software can label a binary quantity as GB, making two displays appear inconsistent even when the byte count matches.

Use the unit reported by the file source if it is defined. For precise reconciliation, use the exact byte size from file properties. A 25 GiB file is about 7.37% larger than a 25 GB file and takes correspondingly longer at the same throughput.

Line speed versus payload throughput

A speed test or fibre plan describes a link under certain conditions, not guaranteed payload for every server. Protocol headers, encryption, acknowledgements and retransmissions consume capacity. Wi-Fi interference, router processing, congestion and server throttling can reduce throughput.

The default 85% is a scenario, not a universal rule. Enter an observed sustained rate divided by line rate when historical data exists. Do not increase efficiency above 100% to force the result to match a cached or compressed download; that is a different data path.

Worked 25 GB example

A decimal 25 GB file contains 200 gigabits. At an ideal 100 Mbps, transfer time is 2000 seconds, or 33 minutes 20 seconds. At 85% effective throughput, the payload rate is 85 Mbps and time rises to about 39 minutes 13 seconds.

The displayed effective MB/s is 85 divided by eight, or 10.625 MB/s. Multiplying that rate by the estimated seconds recovers approximately 25,000 MB. This reverse check confirms that bits, bytes and efficiency were applied in the intended direction.

Wi-Fi and local bottlenecks

A fast internet plan can be limited by Wi-Fi signal, channel congestion, an older device, a 100 Mbps Ethernet port, slow storage or antivirus scanning. Test with a wired capable device near the router before attributing every slow download to the provider.

Mesh backhaul and repeaters can reduce usable throughput. Link-rate numbers shown by Wi-Fi are not application speed. Compare local network transfer, internet speed test and the specific server download separately to locate the bottleneck.

Server and route limits

The remote server may limit each connection or be busy. International routing, peering and long round-trip time affect how quickly transport protocols fill the path. Multiple parallel streams can improve some transfers but can also violate service limits or compete with other users.

A content-delivery network may serve a nearby cached copy at one time and a distant path later. Record date, server and test method when investigating a difference. One fast speed-test host does not prove every destination can deliver that rate.

Other household traffic

Streaming, cloud backup, gaming updates, video calls and other users share connection capacity. The calculator assumes the entered effective rate belongs to this file throughout the transfer. Quality-of-service settings may deliberately prioritise interactive traffic and slow background downloads.

Schedule large transfers outside critical calls or use router controls where appropriate. Pausing other traffic can improve time but does not overcome the server or physical access limit. Monitor data use when the connection has a cap or fair-use policy.

Upload and cloud synchronisation

The same arithmetic applies to upload when speed is replaced with sustained upload Mbps. Many connections are asymmetric, so upload can take much longer than download. Cloud sync may also scan, hash, encrypt and compare files before sending data.

A folder size is not always network payload. Deduplication may send less; version history, metadata or retries may send more. Use provider transfer logs for a completed job when billing or recovery planning depends on exact data volume.

Resume, compression and verification

A resumable download can continue after interruption instead of restarting, but support depends on server and application. Compressed transfer can reduce bytes for text while making little difference for video, photos or archives already compressed. The calculator uses the entered payload size without compression.

After a critical download, verify the published checksum where available. A fast transfer is not useful if the file is corrupt or incomplete. Preserve enough disk space for temporary and extracted files, which can exceed the archive size.

Planning backups and updates

For a large backup or operating-system update, calculate a central time and add a reliability window for variable throughput. Ensure the device remains powered, connected and has storage. Avoid starting a firmware update when load shedding or battery life may interrupt it.

Use the ideal line as a lower bound, not a promise. If the task repeatedly exceeds the efficiency scenario, measure sustained rate over a representative interval and update the input. Capacity planning improves with observed distributions rather than one best speed-test result.

Run a realistic transfer check

Measure sustained throughput over a representative file rather than relying only on a short speed-test peak. Confirm whether the displayed rate is in bits or bytes, note Wi-Fi or mobile signal conditions, and include any server-side cap. For an important upload or download, allow time for verification, retries, encryption and local disk writing after network transfer.