Are there no benchmarks using a simple software, downloading a file from a pc connected with a crossover cable using HTTP, preferrably to the ram-disk?
The reasons for this are:
1. Ftp is a cpu-blog.
2. Samba is a larger cpu-blog.
3. From the results, amiga explorer seems to be a cpu-blog too.
4. A piece of good written software retrieving files using the HTTP-protocol will use little cpu power.
4. Most new 100Mbit switches doesnt handle 10MBit NICs that well, thats why a crossover cable should be used when benchmarking.
5. With the ram-disk you are not affected by software which uses the filesystem in an inappropriate way for fast transfers.
After all, it is the performance of the NIC you want to try to measure, not the performance of your CPUs, right? Then it is a very good idea to try to minimize the CPU-usage!
/Patrik