smartNIC - Page 5
Before I proceed on to the performance of the adapter, here
is a brief smartNIC product feature list :
with Computers with USB Port and Windows98/98SE Operating
USB port. No Additional Power Adapter Required.
Low Cost Alternative
for PCMCIA Cards.
To test the performance of the smartNIC, a 444MB (actual
: 465,985,536 bytes) MPEG movie file was transferred three
times between my smartNIC-desktop combination and the TravelMate
330 in my 10Mbps only home network. No additional devices
were attached to the USB root hub. The TravelMate used the
Xircom Credit Card
10/100 Ethernet Adapter. The systems were rebooted after
every file transfer to flush the cache. The results of the
three transfers were averaged to obtain the mean transfer
In addition, to see how the
smartNIC performed while sharing its USB bandwidth with
other devices, I hooked a Prolink 4-port USB hub to a USB
port on the desktop configuration and connected the HP ScanJet
4200, the Creative WebCamII and the Creative Cobra gamepad
to the hub. Take note that the average theoretical transfer
rate for a port on the USB root hub is 8Mbps, though in
reality, it is lower. The bandwidth available to devices
daisy chained to the port would have to be divided among
the devices. The same 444MB file was transferred while the
WebCamII was left running. Occasional scans were also done.
Again, the results were averaged to obtain the mean transfer
For comparison purposes,
the same 444MB file was transferred between the DFE-530TX-desktop
combination and the Xircom Credit Card-TravelMate 330 combination
in a 10Mbps network and a 100Mbps network to obtain their
respective transfer rates.
From the numbers
above, we can see that daisy chaining USB devices to the
smartNIC does noticeably degrade performance. As expected,
USB network performance as compared to 10Mbps and 100Mbps
networks trails by quite a lot.