Date: Wed, 04 Dec 1996 20:42:40 -0800 From: "Jordan K. Hubbard" <jkh@time.cdrom.com> To: Michael Hancock <michaelh@cet.co.jp> Cc: FreeBSD Hackers <hackers@FreeBSD.org> Subject: Re: New benchmarks to design Message-ID: <5487.849760960@time.cdrom.com> In-Reply-To: Your message of "Thu, 05 Dec 1996 08:53:46 %2B0900." <Pine.SV4.3.95.961204161549.17926B-100000@parkplace.cet.co.jp>
index | next in thread | previous in thread | raw e-mail
> worldStone: cd /usr/src; make world. This is important to people who > build world a lot. In observing, results posted on this list there's a This is actually a useful composite benchmark. However, if you're going to collect worldstone values then you need to be REALLY REALLY sure of several things before you accept the numbers: 1. The /etc/make.conf values are the same on every machine building the world. If I compile with -O2 and -pipe and you just compile with -O and -pipe, our numbers are going to vary widely. 2. The source tree built from is the exact same one, not just a fuzzy value of -current. 3. The source tree is fully cleaned before starting or you will also be adding in the time it takes to remove the old objects when the build starts. 4. The machines have the same amount of memory, for obvious reasons. Jordanhelp
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5487.849760960>