I used Jamal's ttcp-sf (TTCP sendfiled ) also, but no change again. I did'nt understand how sendfile could have helped here. AFAIK sendfile() will just reduce the CPU utilization further as it avoids a user-to-kernel copy. It can increase throughput only in case when it is limited by getting the data to send (an FTP server reading from a file on the hard disk). If without TSO we are getting a healthy 102MBps then thruput should not fall with TSO. Is it possible that in TSO case the kernel is keeping the application provided data longer in a bid to accumulate it to make larger payloads that can then be TSO'ed to the chip. Just a guess ! FYKI, I have also tried with 256KB send buffer size and the results are same. Thanx, tomar On Tue, 12 Aug 2003, David S. Miller wrote: > On Tue, 12 Aug 2003 05:01:21 +0530 (IST) > Nagendra Singh Tomar <nagendra_tomar@adaptec.com> wrote: > > > Any suggestions ? > > Try again with an application that uses sendfile(). > - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
