OK, I've finally run lmbench-1.1 on both machines in "single user" mode
(i.e. with nothing in the way except the portmap daemon, which lmbench
needs to do RPC testing).  NetBSD isn't looking very good yet.  :-(

The first two rows in each table are the ones to compare:

	most.1 is the SunOS-4.1.1_U1 3/260 w/48MB running in single user
	always.2 is the NetBSD-1.1 3/250 w/32MB running in single user

The other two rows are not so important -- I just had the data, so I
thought I'd throw it in to see how minimal multi-user services affect
things:

	most is the SunOS-4.1.1_U1 3/260 w/48MB running in multi-user
	always is the NetBSD-1.1 3/250 w/32MB running in multi-user

Ignore the "file reread".  It's atrocious because the NetBSD box has a
wimpy little old 100 MB Toshiba drive, and the SunOS box has a much
faster Quantum PD 1050S.  Refer to my older postings about filesystem
and raw disk performance to see identical disks compared between
NetBSD-1.1 and SunOS-4.1 (i.e. NetBSD is slower, but in weird ways).

Also note that Mmap re-read is probably far better under NetBSD simply
because NetBSD doesn't go to the trouble of keeping the filesystem
buffer cache in sync as SunOS does.

Oh, yeah, and I have confirmed that the timing loop used by lmbench is
not at all prepared to deal with the mc68020, thus the absurd 9MHz clock
frequency (it's really 25MHz).  I hope the main memory latency is
similarly off, but I won't guarantee it....


                L M B E N C H  1 . 0   S U M M A R Y
                ------------------------------------

                  Comparison to best of the breed
                  -------------------------------

                (Best numbers are starred, i.e., *123)


        Processor, Processes - factor slower than the best
        --------------------------------------------------
Host                 OS  Mhz    Null    Null  Simple /bin/sh  Mmap 2-proc 8-proc
                             Syscall Process Process Process   lat  ctxsw  ctxsw
--------- ------------- ---- ------- ------- ------- ------- ----- ------ ------
most.1    SunOS 4.1.1_U   10    *166  *29.9K *142.2K *231.6K *1368   *109    1.2
always.2     NetBSD 1.1   10     1.0     1.9     2.6     2.5   2.4    2.1   *458
most      SunOS 4.1.1_U   10     1.0     1.1     1.1     1.1   1.1    1.7    1.4
always       NetBSD 1.1   10     1.0     1.9     2.7     3.1   2.4    2.1    1.0

        *Local* Communication latencies - factor slower than the best
        -------------------------------------------------------------
Host                 OS  Pipe       UDP    RPC/     TCP    RPC/
                                            UDP             TCP
--------- ------------- ------- ------- ------- ------- -------
most.1    SunOS 4.1.1_U   *1089   *2227   *4244   *2608   *5336
always.2     NetBSD 1.1     1.2     1.3     1.7     1.4     1.5
most      SunOS 4.1.1_U     1.2     1.1     1.2     1.1     1.0
always       NetBSD 1.1     1.2     1.3     1.7     1.4     1.5

        *Local* Communication bandwidths - percentage of the best
        ---------------------------------------------------------
Host                 OS Pipe  TCP  File   Mmap  Bcopy  Bcopy  Mem   Mem
                                  reread reread (libc) (hand) read write
--------- ------------- ---- ---- ------ ------ ------ ------ ---- -----
most.1    SunOS 4.1.1_U   *1  77%     *2    77%     *3    96%   *7   97%
always.2     NetBSD 1.1  75%   *0    20%     *5    71%    99%  98%    *9
most      SunOS 4.1.1_U  89%  71%    97%    72%    95%    94%  96%   93%
always       NetBSD 1.1  73%   *0    20%    99%    72%     *4  99%   99%

            Memory latencies in nanoseconds - factor slower than the best
                    (WARNING - may not be correct, check graphs)
            -------------------------------------------------------------
Host                 OS   Mhz  L1 $   L2 $    Main mem    Guesses
--------- -------------   ---  ----   ----    --------    -------
most.1    SunOS 4.1.1_U     9   1.0    ???         1.0    No L2 cache?
always.2     NetBSD 1.1     9   1.0    ???        *948    No L2 cache?
most      SunOS 4.1.1_U     9  *137    ???         1.0    No L2 cache?
always       NetBSD 1.1     9   1.0    ???        *948    No L2 cache?


If anyone would like me to run different benchmarks on these systems,
please let me know.  As soon as I can afford another big disk, I'm going
to go to -current and try some kernel profiling....  I should probably
get a faster CPU for my server too if I'm going to be doing all this
stuff!  ;-)

-- 
							Greg A. Woods

+1 416 443-1734			VE3TCP			robohack!woods
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>