Date: Mon, 7 Aug 2000 22:17:31 +0200 From: Francesco Casadei <fcasadei@inwind.it> To: freebsd-questions mailing list <freebsd-questions@freebsd.org> Subject: performance gain of a buildworld using -march option Message-ID: <20000807221731.A241@casimirhost.kasby>
next in thread | raw e-mail | index | archive | help
I have two different machines: a 486 (DX2-50 MHz, 12 MB RAM and
203 MB ide disk) and a 586 (AMD K6-2 3D 300 MHz, 128 MB RAM and 22+
GB SCSI disks), from which I run CVSup to keep up to date with the
4.x-STABLE source tree.
It is worth to make two different buildworld, using -march=i486
for the 486 and -march=pentium for the AMD? What is the performance
gain?
Consider the following solutions: what do you think is better?
1) two different builds:
1.1) make buildworld (and kernel) using -march=i486, then
installworld (and kernel) on the 486;
1.2) make buildworld (and kernel) using -march=pentium, then
installworld (and kernel) on the K6;
2) one buildworld (and kernel) using -march=i486 for both the i486
and the K6
3) the performance gain is so negligible that I shouldn't even bother
to add -march=cpu_type to /etc/make.conf... (i.e. only one
"normal" buildworld, then installworld on the 486 and on the K6)
I appreciate very much your suggestions.
Francesco Casadei
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000807221731.A241>