On Wed, Jul 21, 2010 at 10:47:11PM -0500, Mikus Grinbergs wrote: > > Use of iperf on a local network can be easily reproduced, however. > > I'm sorry - what I am interested in is how quickly I can download a new > build to the XO with my existing setup -- I am NOT interested in how > well a special-purpose tool (iperf) performs on my existing LAN.
Yes, I understand. I've had to subvert my interests in order to be balanced, fair and reproducible in my testing. So that I can isolate the cause of unusual observations. > Anyway - running iperf with an XO-1 as the iperf server: > > My throughput numbers are 1/10 (or even less) of Quozl's. [Even though > the ethernet switch(es) on my LAN are advertised as 100Mb, I've never > gotten high throughput rates on that LAN.] > > By the end of a period of testing, the throughput to the XO-1.5 client > was only 3% slower than the throughput to the XO-1 client. The 'top' > values for (wa) and (hi) were zero for the iperf clients. The only > difference was the (sy) value - 6% for the XO-1.5 client, 2% for the > XO-1 client. Thanks. It sounds like one of the USB to Ethernet adapters has negotiated with the ethernet switch for the lower supported speed of 10Mbit/sec. Given that we've only one data point at 100Mbit/sec ... mine, I don't know this slow data transfer is a problem yet. I'm interested in hearing from others on it. My observations could easily be explained as an interaction between a particular device and the newer XO-1.5 hardware. -- James Cameron http://quozl.linux.org.au/ _______________________________________________ Devel mailing list [email protected] http://lists.laptop.org/listinfo/devel
