Pi 4 speedtest.net much slower than iperf3 over GigE port

[wayne.mac]

I am busy with fast internet tests (over 5G 3GPP) and want to use the Pi4 over remote connection to run throughput test remotely. Unfortunately I have hit a problem that the https://www.speedtest.net/ throughput results are significantly lower that those obtained using iperf3.

[*]- 400Mbps or more using iperf3
[*]- <100Mbps using https://www.speedtest.net/
[*]- 700-945Mbps using iperf direct cable connection

Any help to solve this issue will be greatly appreciated.

[fruitoftheloom]

I am busy with fast internet tests (over 5G) and want to use the Pi4 over remote connection to run throughput test remotely. Unfortunately I have hit a problem that the https://www.speedtest.net/ throughput results are significantly lower that those obtained using iperf3.

[*]- 400Mbps or more using iperf3
[*]- <100Mbps using https://www.speedtest.net/
[*]- 700-945Mbps using iperf direct cable connection

Any help to solve this issue will be greatly appreciated.

Try speedtest-cli the command line version, it is in the repositories ?

This forum has many posts stating Speedtest web page gives inconsistent results....

[wayne.mac]

Yes, sure the results are inconsistent.....But not lower than a factor of 4x compared to a USB boot Linux PC....consistently.

[B.Goode]

I am busy with fast internet tests (over 5G) and want to use the Pi4 over remote connection to run throughput test remotely. Unfortunately I have hit a problem that the https://www.speedtest.net/ throughput results are significantly lower that those obtained using iperf3.

[*]- 400Mbps or more using iperf3
[*]- <100Mbps using https://www.speedtest.net/
[*]- 700-945Mbps using iperf direct cable connection

Any help to solve this issue will be greatly appreciated.

Are you comparing like-with-like here?

Is the remote site being used for the iperf3 test the same one as used by speedtest?

Does 5G denote a WiFi technology, or a 'cellphone' technology?

For most domestic users (things might be different for business or institutional networks) I would expect that 'external' Internet throughput would be lower than the non-routed throughput of an 'internal' device-to-device connection.

My limited understanding of 5G mobile networks is that they are not always faster: speed can be traded for reach and reliability.

[wayne.mac]

Below is the result from iperf3 command and a snapshot of the speedtest result.
Result: Iperf3 receives at 354 Mbps vs 61 Mbps on speedtest.net.

Code: Select all

- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-120.00 sec  16.8 GBytes  1.20 Gbits/sec  0.000 ms  0/13270505 (0%)  sender
[SUM]  0.0-120.0 sec  802670 datagrams received out-of-order
[  5]   0.00-120.00 sec  4.93 GBytes   353 Mbits/sec  0.023 ms  9346325/13270505 (70%)  receiver

[lingon]

Try running the speedtest-cli tool as in this thread https://www.raspberrypi.org/forums/view ... 3&t=251736.