Date: Sun, 28 Dec 2014 22:13:58 +1100 From: Kubilay Kocak <koobs@FreeBSD.org> To: David Chisnall <David.Chisnall@cl.cam.ac.uk> Cc: svn-ports-head@freebsd.org, svn-ports-all@freebsd.org, Gerald Pfeifer <gerald@pfeifer.com>, ports-committers@freebsd.org Subject: Re: svn commit: r375696 - head/multimedia/libx264 Message-ID: <549FE5F6.5010308@FreeBSD.org> In-Reply-To: <6061C442-9051-433E-B347-7F3466234DFD@cl.cam.ac.uk> References: <201412280459.sBS4xbU9021196@svn.freebsd.org> <alpine.LSU.2.11.1412281100080.3428@tuna.site> <549FE2CE.9050006@FreeBSD.org> <6061C442-9051-433E-B347-7F3466234DFD@cl.cam.ac.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
On 28/12/2014 10:05 PM, David Chisnall wrote: > On 28 Dec 2014, at 12:00, Kubilay Kocak <koobs@FreeBSD.org> wrote: > >> (eg: PGO until libprofile_rt in Clang lands in FreeBSD), whether or not >> 4.2.1 provided statistically significant benefits over later versions. > > Shouldn't PGO be something handled by the compiler USES flags, rather than hard-coding it in here and every other port that has a PGO option? > > David > Sounds reasonable. We already have :openmp, so a :pgo doesn't seem too far-fetched. It may or may not matter to the question, but in x264's case, there is also a custom build target that is run for PGO along with the subsequent multi-pass encoding runs. If we could keep any extra PGO logic needed in ports while outsourcing the compiler selection, that sounds like a net gain. Koobs
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?549FE5F6.5010308>