As far as I know X540-T2 out on the market don't do PCI 3.0. Cards I have are PCI 2.1, this means (if I remember my calculations right) this 10G card is caped by PCI bus - 6G max. Basically Intel sells 10G which is caped up to 6G. and this is for the single port. If those ports are both in use, then you'll have to divide this number with 2(avrg. and not precise number).
So, per port on X540-T2, you have maximum 3Gbit/s. in theory, if both ports used and have avrg. the same amount of traffic. if not both - 6Gbit/s Correct me if I'm wrong. //mxb On 9 aug 2013, at 03:35, John Jasen <jja...@realityfailure.org> wrote: > Apologies for the top posting, please. > > Interestingly, despite the E3 you're using being a newer chip, and > having PCIE 3.0, the systems I'm running on Xeon X5570-based CPUs seem > to have a few advantages -- and can push close to 20 Gb in testing > scenarios. > > For example, it looks like the X5570 has better system bus bandwidth and > better memory bus bandwidth (ark.intel.com lets you compare chips side > by side). > > Dunno if that means anything, but its interesting. > > Topping out per 82599 card at ~8k interrupts does not surprise me, as I > was unable to get any of mine beyond that. I personally think the 82598 > is better under OpenBSD, using about 40% of the interrupts for similar > bandwidth. > > The system showing 90% utilization at 16k interrupts surprises me. My > systems showed about 35-40% utilization at 25-30k interrupts. > > You may want to test jumbo frames, just to see what would happen. I > would expect you to see closer to 10 Gb/s with the same number of > interrupts. > > Since I've completely ignored email etiquette tonight, please allow me > to snip through here. > > On 08/08/2013 08:26 PM, Maxim Khitrov wrote: > <snip> >> The BIOS on these firewalls is current. For power-saving options, when >> I first configured these systems I tried turning Intel EIST >> (SpeedStep) off, but this caused OpenBSD to panic during boot. > > My systems are set to maximum performance at all power savings > steppings. I don't know if this is Dell pretending we're all stupid, or > if your BIOS has similar settings. > > <snip> > >> Active Processor Cores: All > > I would turn that off, or at least make it only dual core. > >> As a side note, iperf doesn't crash on FreeBSD when running in UDP >> mode, so I think it's a problem with the OpenBSD package. For these >> tests I stuck with TCP and 1500 MTU. Also, I noticed that a 10 second >> test is not always sufficient to get consistent results, so I'm now >> running all tests for 60 seconds. > > UDP can be a little iffy. FWIW, it never hurts to verify your tool's > results with another tool. I used nuttcp on most of my tests. > > >> That's... a bit faster. The CPU in the desktops is Intel i7-3770, >> which is very similar to the Xeon E3-1275v2. Is this a FreeBSD vs >> OpenBSD difference? > > > Could be. It might be worth testing FreeBSD on your packet forwarding > boxes, just to see if you get similar results. > > -- > -- John Jasen (jja...@realityfailure.org) > -- No one will sorrow for me when I die, because those who would > -- are dead already. -- Lan Mandragoran, The Wheel of Time, New Spring
