Delivered-To: port-sparc-outgoing@NetBSD.ORG
   From: Anders Magnusson <ragge@ludd.luth.se>
   Date: Thu, 24 Oct 1996 11:51:07 +0200 (MET DST)
   Cc: ragge@ludd.luth.se, port-sparc@NetBSD.ORG
   X-Mailer: ELM [version 2.4ME+ PL15 (25)]
   MIME-Version: 1.0
   Content-Type: text/plain; charset=US-ASCII
   Content-Transfer-Encoding: 7bit
   Sender: owner-port-sparc@NetBSD.ORG
   Precedence: list
   Delivered-To: port-sparc@NetBSD.ORG

   > On Wed, 23 Oct 1996, Anders Magnusson wrote:
   > 
   > > > 
   > > Actually, this only shows that there is something fundamentally
   > > wrong with Solaris. It shouldn't be possible to emulate a system
   > > with the gain of double speed.
   > 
   > If I were to guess I'd say that you were probably doing this on systems
   > with a smallish amount of RAM, say 16meg.  Solaris is certainly a memory
   > hog, and doesn't tend to run properly in anything less than 32, with 
   > 40 or more being ideal for most workstations.  You can get by with less
   > if you are not running openwindows or X.
   > 
   It was measured under the same conditions. 32MB internal memory,
   otherwise idling machines and not running X on the same machine.
   (plenty of memory free). I tried to get some interesting info out
   of time, but it only shows that Solaris spends much more time in
   kernel space but _also_ more time in user space (!). This is of
   course dependent of what the system counts in user time.

Quite recently, the linux/sparc people published detailed comparisons of sunos
4.1.4 / solaris 2.5 / linux. In some instances, linux performed twice
as fast as sunos. So i am not surprised learning this also about
netbsd.

Michael