Date: Mon, 11 Feb 2013 14:22:21 -0800 From: Steve Kargl <sgk@troutmask.apl.washington.edu> To: David Chisnall <theraven@FreeBSD.org> Cc: freebsd-current@FreeBSD.org Subject: Re: 7+ days of dogfood Message-ID: <20130211222221.GA83846@troutmask.apl.washington.edu> In-Reply-To: <17E009FB-23FA-4E04-8437-DE81033164DE@FreeBSD.org> References: <20130210000723.GA73630@troutmask.apl.washington.edu> <20130211114811.09e56b55@fabiankeil.de> <17E009FB-23FA-4E04-8437-DE81033164DE@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Feb 11, 2013 at 01:29:12PM +0000, David Chisnall wrote: > On 11 Feb 2013, at 10:48, Fabian Keil <freebsd-listen@fabiankeil.de> wrote: > > > It's unfortunate that the builworld time roughly trippled since > > 2010 but I guess that's progress and a more powerful system > > should fix it. I certainly welcome clang in general, though. > > In that case, it's worth noting that you can shave a fair bit off > the build time by not building gcc. WITHOUT_GCC=yes in src.conf > is worthwhile. > While not building a part of the base system will obviously speed up bulidworld, it seems to me that you're being a little too generous here. The buildworld-speed issue is clearly a problem with clang/llvm. On my very lightly loaded, 4-core opteron system with 16 GB of memory, I see WITH_CLANG="YES" WITH_GCC="YES" rm -rf /usr/obj/* time make -j4 buildworld 3634.55 real 9784.21 user 1286.08 sys WITH_CLANG="YES" WITHOUT_GCC="YES" rm -rf /usr/obj/* time make -j4 buildworld 3489.40 real 9413.90 user 1324.72 sys WITHOUT_CLANG="YES" WITH_GCC="YES" rm -rf /usr/obj/* time make -j4 buildworld 1928.38 real 5254.85 user 1075.68 sys -- Steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20130211222221.GA83846>