--+QahgC5+KEYLbs62
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Tue, Nov 08, 2005 at 06:47:47PM +0100, David Landgren wrote:
> And Kris Kennaway did write:
>=20
> >On Mon, Nov 07, 2005 at 05:15:55PM +0100, Achim Patzner wrote:
> >
> >>Am 07.11.2005 um 16:27 schrieb Ollivier Robert:
> >>
>=20
> [...]
>=20
> >>I remember someone writing that the intermediate state of -CURRENT =20
> >>while removing the giant lock around the kernel wasn't viable with =20
> >>more than four CPUs as it would completely deadlock from time to =20
> >>time. I guess we're a bit further down the road...
> >
> >
> >That certainly seems to be the case. You want to use FreeBSD 6.0,
> >which has excellent performance and stability on SMP in my testing,
> >even with 14 CPUs. FreeBSD 5.4 is definitely not up to it, since VFS
> >is under Giant.
>=20
> Well I'm going to take the plunge and bring an HP Netserver LT 6000 with=
=20
> 6 CPUs up from 5.4-STABLE to 6.
>=20
> Are there any obvious additions or subtractions from a 5.x kernel=20
> configuration file? Is ADAPTIVE_GIANT now obsolete?
I think you still want it - it was a net win the last time I measured
it. That's it though, really.
Kris
--+QahgC5+KEYLbs62
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (FreeBSD)
iD8DBQFDcOXDWry0BWjoQKURApfGAKDnPQIzdmj1qY252sKPY3iPBaM6bgCeJ13x
H3S0WJI/O7laZL7v3rVORSk=
=x5P0
-----END PGP SIGNATURE-----
--+QahgC5+KEYLbs62--