Re: NetBSD 6 is looking pretty good

[David Brownlee <abs%netbsd.org@localhost>]

On 17 October 2012 13:34, Toby Thain <toby%telegraphics.com.au@localhost> wrote:
> On 17/10/12 6:41 AM, John Klos wrote:
>>
>> Hi,
>>
>> I have a VLC which is running netbsd-6 from 1-October. It's been
>> compiling and doing various things for the last two weeks with no
>> problems.
>>
>> I did notice one thing, though. When running benchmarks/dhrystone/, the
>> number I got was:
>>
>> Microseconds for one run through Dhrystone: 569.1
>> Dhrystones per Second: 1757.2
>
>
> I would mention that Dhrystone is an extremely poor integer benchmark. Its
> limitations were known back in the 1980s and I don't think anyone takes it
> seriously.
>
> Here is a much better one (but there are of course many others):
>
>   http://homepages.cwi.nl/~tromp/c4/fhour.html
>
>>
>> When I ran dhrystone on netbsd-5 earlier this year, though, I got 2164
>> which is almost 1.25 times faster. What's also interesting is that the
>> standard number given for dhrystones per second for the VAX 11/780 is
>> exactly 1757, too. Does this mean that the inefficiencies of new
>> toolchains has brought a 25 MHz 1991 machine down to the speed of a 5
>> MHz 1977 machine?
>
>
> That said, I have a 5 MHz MicroVAX II here running NetBSD 1.4.1 if you
> wanted to confirm the latter number. :-)

Might be interesting to see the results from a staticly linked binary
- ideally also a staticly linked NetBSD 1.4.1 binary run on 1.4.1 and
later versions :)