Date: Fri, 31 Mar 2000 18:53:22 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: "Matthew D. Fuller" <fullermd@futuresouth.com> Cc: Matthew Dillon <dillon@apollo.backplane.com>, Andy Farkas <andyf@speednet.com.au>, freebsd-current@FreeBSD.ORG Subject: Re: SMP buildworld times / performance tests Message-ID: <Pine.BSF.4.21.0003311828140.14809-100000@alphplex.bde.org> In-Reply-To: <20000330174453.D10480@futuresouth.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 30 Mar 2000, Matthew D. Fuller wrote:
> Does softupdates provide faster performance than async/noatime? I keep
> /usr/src and /usr/obj as such, would it be faster with softupdates? And
> if so, why?
Async is theoretically faster if it is implemented properly. It is poorly
implemented in FreeBSD, but is good enough for things like makeworld.
From my makeworld logs:
async mounts, standard kernel, standard Makefile.inc1, NOSHARED=yes
--------------------------------------------------------------
>>> elf make world started on Mon Jan 17 12:11:25 EST 2000
>>> elf make world completed on Mon Jan 17 13:23:23 EST 2000
--------------------------------------------------------------
4317.91 real 3237.34 user 636.77 sys
11864 maximum resident set size
1105 average shared memory size
1061 average unshared data size
129 average unshared stack size
9103073 page reclaims
15251 page faults
0 swaps
40230 block input operations
4313 block output operations
0 messages sent
0 messages received
7 signals received
372803 voluntary context switches
198051 involuntary context switches
There were only 4313 disk writes for the whole makeworld. This
shows that -async is doing a good job of stopping writes going to the
disk. 4313 writes is too few too take very long.
Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0003311828140.14809-100000>