Date: Tue, 27 May 2014 23:02:20 -0400 From: "George Neville-Neil" <gnn@neville-neil.com> To: "Navdeep Parhar" <np@FreeBSD.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org, Luigi Rizzo <rizzo@iet.unipi.it> Subject: Re: svn commit: r266757 - in head/sys: conf dev/cxgbe dev/cxgbe/common modules/cxgbe/if_cxgbe Message-ID: <897F102E-BDE2-454B-964A-BAF02369E70E@neville-neil.com> In-Reply-To: <538530E0.9090309@FreeBSD.org> References: <201405271818.s4RIIff9045225@svn.freebsd.org> <CA%2BhQ2%2BgQ5JYUBrPwbSzByCLJ_eExY=VJi161MkC2WnG_5iusNA@mail.gmail.com> <538530E0.9090309@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 27 May 2014, at 20:42, Navdeep Parhar wrote: > On 05/27/14 17:36, Luigi Rizzo wrote: >> >> >> >> On Tue, May 27, 2014 at 8:18 PM, Navdeep Parhar <np@freebsd.org >> <mailto:np@freebsd.org>> wrote: >> >> Author: np >> Date: Tue May 27 18:18:41 2014 >> New Revision: 266757 >> URL: http://svnweb.freebsd.org/changeset/base/266757 >> >> Log: >> cxgbe(4): netmap support for Terminator 5 (T5) based 10G/40G cards. >> Netmap gets its own hardware-assisted virtual interface and won't take >> over or disrupt the "normal" interface in any way. You can use both >> simultaneously. >> >> For kernels with DEV_NETMAP, cxgbe(4) carves out an ncxl<N> interface >> (note the 'n' prefix) in the hardware to accompany each cxl<N> >> interface. These two ifnet's per port share the same wire but really >> are separate interfaces in the hardware and software. Each gets >> its own >> L2 MAC addresses (unicast and multicast), MTU, checksum caps, etc. >> You >> should run netmap on the 'n' interfaces only, that's what they are >> for. >> >> With this, pkt-gen is able to transmit > 45Mpps out of a single >> 40G port >> >> >> and just for the records, this is with 1 core in userland. >> Very cool. >> >> Do you have an easy way to check what kind of throughput >> you get when the netmap interface has one hw queue ? > > Both tx and rx are at around ~33Mpps right now with one hardware queue. > That's more than enough to saturate a 40G link with 128B sized frames. > We need to port more tools to netmap now that it's in more drivers. I'm looking at iperf3 but there are other packet oriented daemons we ought to be porting as well. PTPd is, of course, on my list. Best, George
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?897F102E-BDE2-454B-964A-BAF02369E70E>