On Wed, 2005-07-12 at 01:08 +0100, Robert Olsson wrote: > jamal writes: > > > Results: > > -------- > > > > kernel 2.6.11.7: 446Kpps > > kernel 2.6.14: 452kpps > > kernel 2.6.14 with e1000-6.2.15: 470Kpps > > Kernel 2.6.14 with e1000-6.2.15 but rx copybreak commented out: 460Kpps > > copybreaks help you.. >
Yes, this was bizarre - it is the opposite of what you saw on opterons. I will repeat again the tests tomorrow to be 100% sure. > > And lastly to just play with different prefetch on/off as Robert did. > > Wish we could verify with some realistic traffic load > I should add that I will run something close to what you did to the multi flows later. I just wanted to make sure that i can see some consistency with the simple case. Anything i should tune with routing? I am going to still keep all load on CPU#0. > > I'll noticed this too and turned flow control off too yes. Useless to have, > it makes network debugging very hard. Much better to see packet drops at > the systems thats not able to keep up with load. And especially now with > 6.2.15 when RX-stats counters finally is fixed. Thanks for fixing this. > oh - they are fixed finally ;-> I didnt pay attention. BTW, with flow control another bizarre thing: 50% or so of the times i tried with flow control on, I did get better numbers with flow control than when i turned it off totally;-> But because it is so undertemisnic it is hard to reproduce consistently. It needs some further thought/analysis; it is possible the switch is messing something. cheers, jamal - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html