Subject: Re: New benchmark...
To: None <port-vax@netbsd.org>
From: George Harvey <fr30@dial.pipex.com>
List: port-vax
Date: 01/28/2001 00:22:45
On 26-Jan-01, NetBSD Bob wrote:

[ snip ]

> 
> After lunch now, and it reports:
> 
> Nodes: 385024
> Best time: 1000440 ms
> Nodes per second: 384 (3.7 MIPS)
> 
> Man, that is less that 1/10th the speed of the 486/66 crate, on
> that fellow's web page!
> 

A few more numbers...

MicroVAX 3100/10e, NetBSD 1.4.2, KA41-D, 16MB RAM:
    550822 nodes
    990200 ms
       556 nodes/s
         3.8 MIPS

HP 9000/340, NetBSD 1.4.1, 68030/16.7MHz, 8 MB RAM:
    550822 nodes
   1182813 ms
       465 nodes/s
         3.2 MIPS

Amiga 2000, NetBSD 1.4.1, 68060/50MHz, 64MB RAM:
    550822 nodes
     99381 ms
      5542 nodes/s
        37.7 MIPS

BTW, there is a maximum run-time set within the bench() function in tscp.
The default value is 1000000 ms which means that very slow machines can
timeout without completing all the nodes. The value can be increased by
editing 'max_time' at line 491 in main.c. The 3100 just made it but I had
to increase max_time to get the HP 340 to complete the full test.

Just for comparison, the Dhrystone 2.1 MIPS numbers for the three systems are:

3100:    3.18
340:     3.22
Amiga:  40.21


George