Date: Wed, 29 Feb 2012 14:17:56 -0800 From: Jack Vogel <jfvogel@gmail.com> To: Rafael Ganascim <rganascim@gmail.com> Cc: freebsd-net@freebsd.org Subject: Re: Igb driver - header split feature Message-ID: <CAFOYbc=6LYcKLS2Ckk5KpwmtbyNDaYB_OivrVHz9Yepsfbo%2B1Q@mail.gmail.com> In-Reply-To: <CAD4ZOMwZmDV30s%2BTgQ2eBMBtb10=U3wvTLfF_5nSUHdNpH=i6g@mail.gmail.com> References: <CAD4ZOMwZmDV30s%2BTgQ2eBMBtb10=U3wvTLfF_5nSUHdNpH=i6g@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
It seems to help in some workloads, makes little difference in others, and can even be less performance in yet others. Its just not a feature that is a 100% win, that's why its not on by default. Try it and see. Cheers, Jack On Wed, Feb 29, 2012 at 5:10 AM, Rafael Ganascim <rganascim@gmail.com>wrote: > Hi list, > > I was looking in the igb driver that I use on some Intel nics > (dualport and quadport gigabit) and in the source code, the feature: > > /* > ** Header split causes the packet header to > ** be dma'd to a seperate mbuf from the payload. > ** this can have memory alignment benefits. But > ** another plus is that small packets often fit > ** into the header and thus use no cluster. Its > ** a very workload dependent type feature. > */ > > Somebody is using this feature? What's the results? > > I think that this feature can be helpfull to improve the performance, > allowing the processing of the packet headers more fastest (pfills, > routing, etc). Is this true? > > Thanks in advance. > > Rafael > _______________________________________________ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFOYbc=6LYcKLS2Ckk5KpwmtbyNDaYB_OivrVHz9Yepsfbo%2B1Q>