From owner-freebsd-current  Tue Oct  3 20:01:40 1995
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id UAA14962
          for current-outgoing; Tue, 3 Oct 1995 20:01:40 -0700
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id UAA14956
          for <current@freebsd.org>; Tue, 3 Oct 1995 20:01:28 -0700
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id NAA09750; Wed, 4 Oct 1995 13:00:25 +1000
Date: Wed, 4 Oct 1995 13:00:25 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199510040300.NAA09750@godzilla.zeta.org.au>
To: bde@zeta.org.au, dyson@freefall.freebsd.org
Subject: Re: lmbench 1.1.5 vs 2.2-current
Cc: current@freebsd.org
Sender: owner-current@freebsd.org
Precedence: bulk

>>The faster disk and clustering speed up the file reread benchmark by a
>>factor of 8.

>Yep, it is mostly the clustering, and the larger, on-demand, buffer cache.

Actually it must be almost entirely due to the larger cache.  My system
can't do 8.6MB/sec except out of the cache.  The test only reads up to
8MB.

>...

>>Otherwise, these benchmarks
>>show unexplained lossages (probably mostly due to vfs bloat) of 10-40%
>>for 2.2-current.

>Interesting, the VFS code adds THAT much overhead to the TCP/UDP results???
>(Maybe, but I don't think so)

Some of the networking latency must be for the extra context switch time
caused by calling microtime() twice for every context switch (it should
only be called once).  I don't see how this could account for 100 usec
extra.

Bruce