How to Test Speed Between Two Computers and Find the Cause of Slow Transfers

Testing speed between two computers requires more than copying a file and watching the progress bar. The result can be limited by Wi-Fi signal quality, Ethernet link negotiation, router or switch capacity, disk performance, CPU encryption overhead, background traffic, or the transfer protocol itself. This guide explains how to measure a local connection with iperf3, how to compare network throughput with real file transfers, how to interpret latency and upload or download results, and which optimization steps can address each likely cause. It also shows how to separate an ISP or broadband problem from a local network problem.

Published 2026-11-21 Last updated 2026-11-21 Category: Guides

What Speed Between Two Computers Actually Measures

The speed between two computers is usually a local network measurement, not an internet speed test. A transfer over Ethernet or Wi-Fi travels through network adapters, cables, switches, routers, and storage devices. By contrast, a broadband test measures the path between your device and a remote test server through your ISP, modem, and the wider internet.

Use throughput to describe the amount of data transferred per second, latency to describe delay, and packet loss to describe missing data that must be retransmitted. Network tools normally report megabits per second, while file managers often show megabytes per second. Since one byte contains eight bits, 800 Mbps is theoretically equal to 100 MB/s before protocol overhead and hardware limitations.

How to Test the Connection with iperf3

iperf3 is a practical way to test network capacity without involving the file systems on either computer. Install it on both devices, connect them to the same network, and use one computer as the server.

  1. On the first computer, run iperf3 -s.
  2. On the second computer, run iperf3 -c SERVER_IP, replacing SERVER_IP with the first computer's local address.
  3. Run the test several times and compare the results.
  4. For the reverse direction, use iperf3 -c SERVER_IP -R.

The normal test shows traffic from the client to the server, while the reverse test helps identify whether upload and download performance differs. Run the test near the router, then repeat it from the normal working location. A large difference usually points to Wi-Fi coverage, interference, or a client device limitation.

Common Causes of Slow Transfers

Wi-Fi signal quality and interference

Weak or congested Wi-Fi is a common reason for inconsistent speed between computers. Distance, walls, neighboring networks, Bluetooth devices, and busy 2.4 GHz channels can reduce the actual throughput far below the advertised wireless link rate. Check the connection at different locations, compare the 5 GHz or 6 GHz band with 2.4 GHz where supported, and test with both computers connected by Ethernet.

Ethernet link negotiation

A negotiated 100 Mbps link can limit a connection that should support 1 Gbps or more. Damaged cables, old network adapters, loose connectors, or a switch port with a problem may cause the devices to fall back to a slower mode. Check the link speed shown by each operating system and network adapter. Replace the cable and test another router or switch port if the reported speed is lower than expected.

Router or switch limitations

Network equipment capacity can become the bottleneck when traffic passes through an older router, an entry-level switch, or a device performing heavy firewall, VPN, or traffic inspection work. Test the computers directly through a capable switch when possible, then compare that result with the normal path. If direct testing is faster, review the router's port speeds, firmware, CPU load, and enabled security features.

Slow storage on one computer

Disk performance can make a healthy network appear slow during file copying. Hard disk drives, nearly full SSDs, USB storage, and drives handling other tasks may not write data quickly enough. Use iperf3 to measure the network independently, then compare it with a transfer between fast local storage locations. If network throughput is high but file copying is slow, inspect disk utilization and test with a smaller file set on a different drive.

Small files and file system overhead

Many small files usually transfer more slowly than one large file with the same total size. Each file requires metadata operations, permission checks, directory updates, and protocol requests. To isolate this cause, test one large file and a folder containing many small files. If the large file performs well while the folder is slow, create an archive for the transfer or use a tool that can parallelize operations carefully.

Protocol, encryption, or sharing settings

The transfer protocol can limit performance through encryption, authentication, compression, or compatibility settings. SMB shares, SFTP, VPN tunnels, cloud synchronization tools, and remote desktop redirection may produce different results on the same network. Compare a raw iperf3 test with an SMB or SFTP transfer. High CPU usage during the slower test may indicate encryption or compression overhead rather than a cable or ISP problem.

Background traffic and device workload

Competing traffic from cloud backup, operating system updates, video streaming, security scans, or another user can reduce available bandwidth. CPU, memory, and disk activity on either computer can also affect the result. Repeat the test when both devices are idle, pause synchronization temporarily, and check resource usage while the transfer is running.

Duplex mismatch, packet loss, or driver problems

Duplex errors and packet loss can cause retransmissions, low throughput, and unstable results even when the reported link speed looks normal. Inspect interface statistics for errors, dropped packets, or retransmissions. Update network adapter drivers and router firmware, replace suspect cables, and test with a different adapter. A continuous ping can help reveal intermittent loss, but it does not replace a throughput test.

How to Interpret the Results

First compare the negotiated link rate with the iperf3 result. A 1 Gbps Ethernet link will not normally deliver a full 1,000 Mbps because of protocol overhead, operating system behavior, and hardware limits, but a result close to 100 Mbps may indicate a 100 Mbps link or another bottleneck. Wireless rates are especially variable, so compare repeated tests rather than relying on the connection icon.

Next compare both directions. If the client-to-server result is much slower than the reverse result, investigate the sending computer, its upload path, storage, or adapter settings. If both directions are low only on Wi-Fi, focus on signal quality and interference. If iperf3 is fast but file transfers are slow, investigate storage, protocol overhead, permissions, and small-file behavior.

Latency is not the same as bandwidth. A low-latency connection may still have poor throughput, while a high-throughput connection may show noticeable delay under load. Run a continuous ping during the transfer to check whether latency rises sharply. This can indicate queueing or bufferbloat in the router or wireless network.

Optimization Steps That Usually Help

  • Use Cat5e or better Ethernet cabling for gigabit connections and replace damaged cables.
  • Connect both computers to the same capable gigabit or multi-gigabit switch when practical.
  • Move Wi-Fi access points to an open, central location and select a less congested channel.
  • Prefer 5 GHz or 6 GHz Wi-Fi for short, high-throughput connections when supported by both devices.
  • Update router firmware, network adapter drivers, and operating system networking components.
  • Pause background backups and updates while measuring performance.
  • Test with one large file before evaluating folders containing many small files.
  • Use iperf3 to separate network performance from disk and file-sharing performance.
  • Check router quality-of-service settings if latency increases sharply during transfers.

When the ISP Is or Is Not the Problem

A local computer-to-computer test does not directly measure your ISP, fiber, cable broadband, or modem service. If iperf3 is fast on the local network but internet downloads or uploads are slow, run a broadband test to a nearby server and compare the result with your subscribed service conditions. Check the modem's signal status and test over Ethernet before concluding that Wi-Fi is responsible.

If both local transfers and internet tests are slow, begin with cables, adapters, router ports, and computer workload. If local performance is normal while only remote services are slow, the cause may be the ISP path, a remote server, congestion, or the application protocol. Recording test time, direction, connection type, negotiated link rate, latency, and packet loss makes it easier to identify a repeatable pattern.

A Reliable Testing Checklist

  1. Record whether each computer uses Ethernet or Wi-Fi.
  2. Check the negotiated link speed on both network adapters.
  3. Run iperf3 in both directions at least three times.
  4. Repeat the test with background traffic paused.
  5. Compare a large-file transfer with a folder of small files.
  6. Monitor CPU, memory, disk activity, latency, and packet loss.
  7. Change one component at a time, such as the cable, port, Wi-Fi band, or protocol.
  8. Run an internet speed test separately if broadband performance is also in question.

For additional broadband diagnostics, compare local measurements with a reputable internet speed test, but keep the two types of results separate. This prevents a local network bottleneck from being mistaken for an ISP problem.