On 7/20/07, Ståle Kristoffersen <staale_at_kristoffersen.ws> wrote: > On 2007-07-20 at 08:10, Jack Vogel wrote: > > On 7/20/07, Ståle Kristoffersen <staalebk_at_ifi.uio.no> wrote: > > >On 2007-07-20 at 09:08, Vince wrote: > > >> Silly as it might seem when did you last recompile the iperf port? I was > > >> about to post a me too post yesterday, until I updated my system to use > > >> the latest sched_ule at which point iperf froze my system completely, > > >> recompiling the port fixed the freeze and the <100Mbit/sec on localhost. > > >> I think it was compiled against the wrong threading library (been a > > >> while since i last recompiled it.) > > > > > >Actually I just spotted this myself, when I recompiled I got in the range > > >of 600Mbits/sec to localhost. However I do noot see the increase you see > > >when running it multiple times. I still think 600Mbits/sec is a bit low. > > > > > >If I run it to a remote host I see the same numbers (around 600Mbits/sec) > > > > I am actually quite impressed with the stack right now, with our new > > Oplin hardware and the ixgbe driver I have 9.8 Gb/s pumping thru > > it. Of course its all about adequate hardware, configuration, and > > tuning. > > Then it must be something I'm doing wrong. It's not that I'm running such a > beast of a machine, but it should manage to fill a gigabitlink (and it did > a few months ago!). > > Can you point out any tuning I might have detuned? The stack being capable, as I said, of delivering the goods says nothing of the driver, what NIC is it? Did you ask if anyone else with that hardware is seeing a problem, I came into this thread somewhat late. JackReceived on Fri Jul 20 2007 - 17:03:51 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:15 UTC