Short answerHigh speed does not guarantee smooth latency. First compare HTTP RTT without load and during data transfer, then compare this with in-game ping.

What's the difference between speed and responsiveness?

Velocity describes the amount of data per time. The moment when a small message reaches the other side is also important for the game. While someone is sending video to the cloud, a queue may accumulate on the network equipment. The connection remains fast, but individual actions receive a response later.

The test below creates a limited download and upload of data while simultaneously measuring the latency to our node. It consumes traffic: on a mobile plan, start with a small limit. Do not run a second speed test at the same time.

Bufferbloat test

Free

Loading the check...

How to compare results

The test first collects background samples, then measures transmission latency. Look at the difference in medians and at the graph: one surge and constant growth are different situations. If a fast channel managed to transmit the entire permitted volume in an instant, there may be little data under load. This result is designated as limited; repeat with a larger limit.

Write down three conditions: cable or Wi-Fi, whether VPN is enabled, who else is using the network. Repeat the measurement after a pause in cloud synchronization. If the delay only changed after stopping it, you have a reproducible script for further tuning.

What to do if the delay increases

  1. Repeat the cable test. The radio channel can add its own delays, especially in a busy band.
  2. Find applications with constant outgoing transfer: backup, cloud drive, video sending, torrent client.
  3. See if the documentation for your particular router model includes SQM or QoS queue management. The names are similar, but the functions of the manufacturers are different.
  4. After changing one parameter, repeat the same test and compare the result. Save your original settings so you can revert them.

Don't buy a more expensive plan based on this measurement alone. If the restriction is related to queues or Wi-Fi, changing the tariff will not necessarily eliminate the cause.

When the other route is to blame

Our HTTP RTT is the delay to the Calcomera node. The game server is located in a different location, and the game client measures the connection in its own way. So 20ms here and 70ms in game can both be valid observations.

If everything is fine under load, check lost messages and stability. Then compare regions within the game if it allows you to select a server. Save the match time and network game schedule: they are more useful than trying to call any jump a “bufferbloat”.

Why the test does not have a universal grade of A+

The result depends on the distance to the node, load duration, background activity and server capabilities. The initial values ​​and conditions are shown here. They help you compare your network before and after a change; the promise of a single letter to describe any game would be too strong.

Sources and methodology

Calcomera editorial teamTests, explanations and open limitations of the method. About the editors