>> so, jack, apparently drivers have imrpvoed =). these numbers
>> were acheived on PII's, intel FX/BX chipsets.
>
>I don't think your testing substantiates that claim; raw tcp transfer
why don't you think that? i used 'ttcp', as i mentioned, to test raw
tcp transfer speeds.
>max is 94mbit. A good 100mbit card and drivr should be able to get this
>as a sustained transfer rate with netperf. Using higher level protocols
wow, amazingly enough, netperf produced exactly the same results as ttcp.
why do you like netperf better?
>(nfs, ftp) aren't really valid, since those programs have some overhead
obviously. i was giving my realworld performance statistics, because
frankly that's what drno's top speed report resembled, so i was giving
some points for possible comparison. i am not a dolt...
>beyond bare tcp or udp. Sorry to be a nitpicker, but I think it is valid
>to get down and dirty with raw tcp/udp when evaluating network hardware
>and drivers.
so, did you just misread what i said the first time (tests with ttcp
giving around 94mbit/s), or can you tell me why netperf's results are
more valid than ttcp's?
---------------------------------------------------------------------------
Send administrative requests to [EMAIL PROTECTED]
