[ECOS] TCP/IP test results: nc_test vs sink/source

Grant Edwards grante@visi.com
Tue Apr 25 11:46:00 GMT 2000

I've noticed something odd when comparing the results of the
tcp tests...

The nc_test_master/nc_test_slave test reports TCP
echo throughput of 496K bytes/second (no load), and UDP echo
throughput of about 660K bytes/second (also with no load).

The tcp_source/tcp_echo/tcp_sink test always reports exactly
50K Bytes/second.  Same Linux host is used for nc_master,
tcp_source, tcp_sink.  This translates into two 10ms clock
ticks per 8k buffer.

I was wondring -- is this 50.0K Bytes/second limit due to the test
programs, the Linux stack, the eCos stack, or do I need to look
more closely at my ethernet driver?

Grant Edwards

More information about the Ecos-discuss mailing list