Subject: Re: SS/10 SMP and NetBSD - CURRENT
To: matthew green <mrg@eterna.com.au>
From: Josh Paetzel <friar_josh@webwarrior.net>
List: port-sparc
Date: 03/31/2003 08:08:02
On Mon, Mar 31, 2003 at 01:39:29PM +1000, matthew green wrote:
>
> On Sun, Mar 30, 2003 at 08:32:00PM -0600, Josh Paetzel wrote:
> >
> > So any thoughts on what would give me better performance, 1 cpu with cache
> > enabled or 2 cpus without cache? The box is going to be used as a headless
> > web/mail server with fairly low loads on it.
> >
>
> Very hard to say, you will be able to run more processes at once but
> they will run slower. It really depends on whether or not you will
> notice the slow down in a single thread or not. Probably your best
> bet would be to suck it and see, if it is perceptibly slower just
> reboot to a uniprocessor kernel.
>
> i noticed about a 30% slowdown with the cache disabled... so 2 cpus
> might be faster if both are running...
>
>
> .mrg.
Sounds like I'll be better off with one cpu then, but I've always believed in
trying out all of my options. I found your patch to cache.c in the mailing
list archives, but it fails to apply cleanly.
===root@www ('tty') /storage/src/sys/arch/sparc/sparc -> patch <patchfile
Hmm... Looks like a new-style context diff to me...
The text leading up to this was:
--------------------------
|Index: cache.c
|===================================================================
|RCS file: /cvsroot/src/sys/arch/sparc/sparc/cache.c,v
|retrieving revision 1.77
|diff -p -r1.77 cache.c
|*** cache.c20 Jan 2003 22:15:54 -00001.77
|--- cache.c4 Feb 2003 00:10:46 -0000
--------------------------
Patching file cache.c using Plan A...
Hunk #1 succeeded at 113 with fuzz 2 (offset -111 lines).
Hunk #2 failed at 127.
1 out of 2 hunks failed--saving rejects to cache.c.rej
done
Any thoughts on this? I'm willing to offer you ssh and root on this box if
it's a matter of you needing a HyperSparc box to work things out on.
Thanks again for your help.
Josh Paetzel