Subject: Re: -RS/6000
To: None <port-powerpc@netbsd.org>
From: Kamal R Prasad <kamalrpr@in.ibm.com>
List: port-powerpc
Date: 03/28/2003 05:08:44
can some one tell me what is the effort reqd to port netbsd onto POWER=
based systems?
are there end-users who would like to see that happen?
thanks
-kamal
----- Forwarded by Kamal R Prasad/India/IBM on 03/28/2003 10:36 AM ----=
-
=
=20
Jochen Kunz =
=20
<jkunz@unixag-kl. To: Kamal R Prasad=
/India/IBM@IBMIN =20
fh-kl.de> cc: =
=20
Subject: Re: -RS/6000 =
=20
03/27/2003 06:13 =
=20
PM =
=20
=
=20
=
=20
On 2003.03.27 14:10 Kamal R Prasad wrote:
> I have access to 43P-100
NetBSD should run on this. (The installer does not work correct, but
once the system is installed it is very stable.)
> as well as POWER based systems.
> what is the effort reqd to port it to POWER based systems?
I don't know enough about POWER and its relation to PowerPC. This can b=
e
everything from fast and simple to complex and very time consuming. Ask=
on port-powerpc@netbsd.org. There are the gurus who have the knowledge
to give you a better answer.
> will it find any end-users?
I am sure it will. There are many RS6k machines out there and many
people who like NetBSD.
> how mnay processors does netbsd scale to?
I think it has the same limits as every SMP implementation with a big
kernel lock. So, depending on the task you have to do, the practical
limit is from 4 to 8. It may run on more CPUs, but this would be
inefficient. This is no limit in NetBSD, it is a systematic limit in SM=
P
systems. You need a different architecture for more CPUs. Somthing like=
a cluster (RS/6000 SP, VAX/Alpha with OpenVMS or Tru64-Unix,
Linux-MOSIX) or NUMA (Sequent Dynix, SGI IRIX).
--
tsch=FC=DF,
Jochen
Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/
=