Date: Fri, 27 Feb 2015 08:46:25 +0000 From: Alexey Dokuchaev <danfe@FreeBSD.org> To: Andrew Rybchenko <arybchik@FreeBSD.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r279351 - head/sys/dev/sfxge Message-ID: <20150227084625.GA72768@FreeBSD.org> In-Reply-To: <201502270739.t1R7dAiK016165@svn.freebsd.org> References: <201502270739.t1R7dAiK016165@svn.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Feb 27, 2015 at 07:39:10AM +0000, Andrew Rybchenko wrote: > New Revision: 279351 > URL: https://svnweb.freebsd.org/changeset/base/279351 > > Log: > sfxge: expect required init_state on data path and in periodic calls > > With the patch applied the number of instruction events is 1% less and > number of mispredicted branch events is 5% less under multistream TCP > traffic load close to line rate. Yuck. This brings more Linux' stink to our code. Why compiler can't do this itself? From what I see, __predict_false was added to != checks, and __predict_true for == ones. Looks like a pattern to me... ./danfe
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150227084625.GA72768>