On Tue, 21 Jun 2016 23:16:52 -0700 Eric Dumazet <eduma...@google.com> wrote:
> When qdisc bulk dequeue was added in linux-3.18 (commit > 5772e9a3463b "qdisc: bulk dequeue support for qdiscs > with TCQ_F_ONETXQUEUE"), it was constrained to some > specific qdiscs. Thanks for extending this! > With some extra care, we can extend this to all qdiscs, > so that typical traffic shaping solutions can benefit from > small batches (8 packets in this patch). I'm fine with limiting this to 8 packets (xmit_more), as that seem to be the minimum needed TX batch size (according to Luigi's original netmap article[1] figure 7) for doing 10G wirespeed. [1] http://info.iet.unipi.it/~luigi/papers/20120503-netmap-atc12.pdf > For example, HTB is often used on some multi queue device. > And bonding/team are multi queue devices... > > Idea is to bulk-dequeue packets mapping to the same transmit queue. > > This brings between 35 and 80 % performance increase in HTB setup > under pressure on a bonding setup : > > 1) NUMA node contention : 610,000 pps -> 1,110,000 pps > 2) No node contention : 1,380,000 pps -> 1,930,000 pps > > Now we should work to add batches on the enqueue() side ;) Yes, please! :-))) That will be the next big step! > Signed-off-by: Eric Dumazet <eduma...@google.com> > Cc: John Fastabend <john.r.fastab...@intel.com> > Cc: Jesper Dangaard Brouer <bro...@redhat.com> > Cc: Hannes Frederic Sowa <han...@stressinduktion.org> > Cc: Florian Westphal <f...@strlen.de> > Cc: Daniel Borkmann <dan...@iogearbox.net> Acked-by: Jesper Dangaard Brouer <bro...@redhat.com> This is great stuff! -- Best regards, Jesper Dangaard Brouer MSc.CS, Principal Kernel Engineer at Red Hat Author of http://www.iptv-analyzer.org LinkedIn: http://www.linkedin.com/in/brouer
