TinyRex Ethernet Performance

[kwebster83]

Hi,

I'm just wondering whether any performance testing has been done on the TinyRex gigabit ethernet?

I'd be really interested to see what the maximum throughput achieved was and over what length of cable.

Regards,
Kevin

[robertferanec]

Hi Kevin,

I do not have the perfect setup, but I hope this will help you. This is what I run this morning. OpenRex is similar to TinyRex (they both use iMX6).

Connection: My computer (virtual machine 192.168.0.245) -> 2m cable -> Mikrotik router -> 2m cable -> OpenRex (192.168.0.251)

Note: There may be some limitations on my virtual machine settings.

Server receives / OpenRex sends

[email protected]:~/iperf# iperf3 -c 192.168.0.245 -t 60 -i 10
Connecting to host 192.168.0.245, port 5201
[  4] local 192.168.0.251 port 51350 connected to 192.168.0.245 port 5201
[ ID] Interval           Transfer     Bandwidth       Retr  Cwnd
[  4]   0.00-10.00  sec   509 MBytes   427 Mbits/sec  192    918 KBytes
[  4]  10.00-20.00  sec   505 MBytes   424 Mbits/sec  177    551 KBytes
[  4]  20.00-30.00  sec   502 MBytes   422 Mbits/sec   20    378 KBytes
[  4]  30.00-40.00  sec   501 MBytes   420 Mbits/sec    1    433 KBytes
[  4]  40.00-50.00  sec   499 MBytes   418 Mbits/sec   31    434 KBytes
[  4]  50.00-60.00  sec   495 MBytes   415 Mbits/sec    0    574 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-60.00  sec  2.94 GBytes   421 Mbits/sec  421             sender
[  4]   0.00-60.00  sec  2.94 GBytes   421 Mbits/sec                  receiver

iperf Done.
[email protected]:~/iperf#

OpenRex receives / Server sends

[email protected]:~/iperf# iperf3 -c 192.168.0.245 -t 60 -i 10 -R
Connecting to host 192.168.0.245, port 5201
Reverse mode, remote host 192.168.0.245 is sending
[  4] local 192.168.0.251 port 51354 connected to 192.168.0.245 port 5201
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-10.00  sec   146 MBytes   123 Mbits/sec
[  4]  10.00-20.00  sec   143 MBytes   120 Mbits/sec
[  4]  20.00-30.00  sec   142 MBytes   119 Mbits/sec
[  4]  30.00-40.00  sec   140 MBytes   118 Mbits/sec
[  4]  40.00-50.00  sec   144 MBytes   121 Mbits/sec
[  4]  50.00-60.00  sec   144 MBytes   121 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-60.00  sec   860 MBytes   120 Mbits/sec  27108             sender
[  4]   0.00-60.00  sec   860 MBytes   120 Mbits/sec                  receiver

iperf Done.
[email protected]:~/iperf#

Some time ago we also tested PCIE Ethernet:
http://www.imx6rex.com/software/imx6...on-tests/#pcie

and also a Wifi card (but performance depends on distance and antenna):
http://www.imx6rex.com/software/imx6...sts/#wifi_4965

[kwebster83]

Hi Robert,

Thanks for the figures. So I guess roughly speaking, I'd expect to get about 420Mbit send, 120Mbit receive.

I wonder why there is such a difference between the send and receive? Also, the re-transmissions for the first couple of OpenRex sends was quite high - do you know if this is always the case?

I'm looking at integrating TinyRex as a server in an embedded application which primarily has ethernet devices (a couple of very high data rate sensors) so hence I'm keen to know how the module performs.

Kind regards,
Kevin

[robertferanec]

Yes, I noticed that difference between transmit / receive too. That is interesting.

The numbers may be influenced by my setup. It was a very quick test setup e.g. I am not really sure how the router is set and also using virtual machine as a server is not probably the best (virtual machine could cause re-transmissions). Maybe the best would be try two boards opposite of each other running nothing else just the iperf, but that would take me more time to setup and play with it. When I have some time in coming weeks, I could try it. With this quick answer I only wanted to give you an idea about the numbers - I hope, that helped a little bit too.

[kwebster83]

Hi Robert,

Your information has been very helpful, thank you. It gives me a very good idea of the baselines that will be achieved and also gives me hope that a lot more performance can be gained - especially on the receive side.

I'd be very keen to hear of any TinyRex to TinyRex performance testing as this is something that I'm also looking at. Almost like a master-slave type setup - maybe with a quad core as the master and a solo/dual as the slaves.

Kind regards,
Kevin

[robertferanec]

I will double check with Pavol from VOIPAC if they can try it. I do not have TinyRex here (I moved to US and took only some boards - mostly OpenRex). So, shall I ask them to try "TinyRex QUAD <-> TinyRex SOLO", direct connection, iperf application, transfer both directions?

[kwebster83]

Hi Robert,

Those sound like perfect tests, thank you.

Kind regards,
Kevin