Subject: Re: benchmarks... (was: Re: More on UFS performance)
To: None <>
From: David S. Miller <>
List: current-users
Date: 12/01/1994 21:01:47
   X-Authentication-Warning: Host localhost didn't use HELO protocol
   Date: Fri, 02 Dec 1994 10:04:10 +1100
   From: matthew green <>
   Precedence: list
   X-Loop: current-users@NetBSD.ORG

   i fully agree with this;  if netbsd is going to "win" it needs to perform.
   as much as i dislike solaris 2.x, sun have done a hell of a good job in
   increasing it's performance in 2.4 -- they claim it's faster than 4.1.x
   in all areas now (not that i believe them, but it could be possible).

Yes, but understand that reliability in the 2.x solaris versions goes
down at an equal ratio as the increased performance. NetBSD on the
sparc at least doesn't have bizarre NFS bugs at which a solaris sun4d
server decides that it will talk to 3 hosts and nfs serve to them and
not it's 120 other clients. This is a serious problem and solaris has
it among others, and the object file patches they release ugh! I don't
trust any kernel thread and/or SMP implementation that I can't take a
look at before throwing it onto a production scheme, but the current
industry makes it such that this is how it has to be done.

Convince corporate america that if they put netbsd and linux on a slew
of P90's they will get better performance and new levels of
reliability and support that they've never seen before (and free too!)
then we won't have to wine about these things. Commercial unix
software vendor have to start recognizing us as a platform to port to,
this will help the above. I want to see a day where I never ever
install an OS that I can't look at the source to without signing away
my ability to write free code on that machine for the next 7 years,
this is complete crapola.

David S. Miller