Date: Mon, 28 Nov 2005 06:03:52 -0500 From: Kris Kennaway <kris@obsecurity.org> To: Kris Kennaway <kris@obsecurity.org> Cc: amd64@FreeBSD.org, Ruslan Ermilov <ru@FreeBSD.org> Subject: Re: My make buildworld FreeBSD 6.0R amd64 test Message-ID: <20051128110351.GA16593@xor.obsecurity.org> In-Reply-To: <20051128105215.GA16457@xor.obsecurity.org> References: <438AA21E.2030702@gneto.com> <20051128104253.GB59977@ip.net.ua> <20051128105215.GA16457@xor.obsecurity.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--tThc/1wpZn/ma/RB Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Nov 28, 2005 at 05:52:16AM -0500, Kris Kennaway wrote: > On Mon, Nov 28, 2005 at 12:42:53PM +0200, Ruslan Ermilov wrote: > > On Mon, Nov 28, 2005 at 07:22:22AM +0100, Martin Nilsson wrote: > > > Hi list, > > >=20 > > > I made some simple comparing of CPU performance between intel and AMD= =20 > > > CPU:s on FreeBSD 6.0R The intent was to see how things like dualcore = and=20 > > > Hyperthreading affects a rather CPU bound real world job. > > >=20 > > > The results can be seen here: > > > http://www.mullet.se/support/amd64-buildworld.html > > >=20 > > > Please don't publish this anywhere else, I want to have your comments= =20 > > > and a chance to fix any errors before I get slashdotted and flamed to= =20 > > > hell :-) > > >=20 > > For the starters, you're compiling two different sets of files, > > one for i386 and another for amd64. You'd better compare the > > same set. This can be done (approximately) by doing a cross-build > > for say i386: "make -jN buildworld TARGET_ARCH=3Di386". Also, given > > the amount of memory that you have in these machines, it shouldn't > > be a problem to put /usr/src and /usr/obj onto a memory disk. >=20 > That will only cross-compile on amd64; isn't it better to > cross-compile a third party (e.g. sparc64) on both i386 and amd64? Actually, I don't think it matters..the compilers are still different on amd64 and i386, so you're comparing two different binaries and you don't know how different their performance is. It'll only be meaningful for comparisons within the same architecture (amd64 or i386), so you might as well not cross-build. Kris --tThc/1wpZn/ma/RB Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (FreeBSD) iD8DBQFDiuQVWry0BWjoQKURAk5DAKCC1RyQzFOa8lJefhYRSaqFXJmKUQCg360U a2O5AVHyOHbznLQJg9fYFtw= =Tayz -----END PGP SIGNATURE----- --tThc/1wpZn/ma/RB--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051128110351.GA16593>