real-world throughput using TCP/IP?

[MAugustis]

I have been running some tests on TCP throughput (TCP NOT UDP). I am trying to
figure out what to expect for the highest real-world transfer throughput. Using
two 3Ghz desktop PCs running XP pro, file transfer speed seems to top out
around 7Mbps. That's only 880kBps! On a 100Mb dedicated LAN (nobody else on
it), I would expect to get better than that.

Is it the protocol itself, the network stack speed, the media, system setup or
what that gets me less than 10% of the theoretical throughput?

Thanks!

 

[Rick Wintjen]

I have been running some tests on TCP throughput (TCP NOT UDP). I am trying to
figure out what to expect for the highest real-world transfer throughput. Using
two 3Ghz desktop PCs running XP pro, file transfer speed seems to top out
around 7Mbps. That's only 880kBps! On a 100Mb dedicated LAN (nobody else on
it), I would expect to get better than that.

Is it the protocol itself, the network stack speed, the media, system setup or
what that gets me less than 10% of the theoretical throughput?

Thanks!

What type of program are you using to transfer the files, Windows
Explorer, ftp, ?

 

[MAugustis]

It doesn't seem to matter what is used. I have tried explorer as well as a
simple sockets program I wrote for both ends.

I have finally tried taking the router out of the pickture. Using a crossover
cable (direct connetion)

I go from 7Mbps to 6 MBps !! Yes that's bits to bytes.