[Shootout-list] X per second scoring system

Bengt Kleberg bengt.kleberg@ericsson.com
Thu, 30 Sep 2004 13:26:09 +0200


Brandon J. Van Every wrote:
> Bengt Kleberg wrote:
...deleted
>>
>>ok, you may prefer 2 valued logic. personally i prefer shades-of-gray.
> 
> 
> No, I like shades of grey just fine.  This is not one of them.


how about if you pretend this is a shades-of-gray area? it would be to 
your liking, since you like such an area. it would be to my liking since 
i think it is a shades-of-gray area. we would both be better of. ok?


>  As I
> said, if you really care about readability *that* much, comment and
> offset the boilerplate properly.
> 
> //======================================================================
> ======
> // LIKE THIS
> //======================================================================
> ======
> 
> If you're not willing to add such comments, then you're not sufficiently
> serious about readability.  Thus not to be taken so seriously about what
> would be 'harmful' to add to the tests, from a readability standpoint.

so, unless i am willing to add boilerplate comments to all the tests in 
the shootout, i am not serious whan i think it would be a (minor) 
obfuscation to add timers to the test?

i think we can agree to disagree on this particular subject.

...deleted

> 
> Yep.  I assume the 10 C FFIs I've looked at in my life, and the C
> standard, and the actual number of arguments to an OS timer function
> call, are representative of the difficulty level of the problem.  You're
> making a theoretical mountain out of a factual molehill.

this is correct. i tend to do that when confronted with exhausted 
researches that does not exhaust.
i much prefer the number 10, or a more humble ''some'', in this case.


>>what is it that stops us from using 30..60 as the value to replace
>>''somehting elkse''?
> 
> 
> Probably something to do with human sense of time, and how we end up
> designing OSes and machines in practice, particularly with regards to
> benchmarking.

so if i say that the test could be run for 1, 2 or something else 
seconds, it is the human sense of time that stops us from having a value 
of 30 (or up to 60) instead of 1 or 2?

this is surprising. i would not have expected that.

how should i formulate my willingness to let the test take 1, 2 or 
something else seconds, so that it is possible for those with a human 
sense of time to understand that i am willing to use other number of 
seconds than 1 or 2? say 30 (up to 60) seconds?


> 
>>however, we are (well, i
>>am) talking about ''timing trials'' from where the quoute
>>came. and yes,
>>if you look in the shootout you will find just such a graph.
> 
> 
> I am thick, I think you should provide me a URL of such a graph.

now, i would not expect somebody that has a tone of writing that implies 
that he knows _the-answer_, to actually say that he has not looked 
thoroghly at what he is talking about. such a thing would make me 
unpleasantly surprised. i do not like beeing unpleasantly surprised. it 
is silly of me, but i know that already.


...deleted
>>however, we are (well, i am) talking about ''timing trials''
>>where there
>>is a graph for every test. you have read the paper, yes?
> 
> 
> I'm sorry, I have done a 'famous name transposition'.  I meant
> Kernighan.
> 
> Those K people.  sheesh.
> 

i take this tio mean that you have read the paper. i  would then like it 
if you could explain the difference between the shotout graphs and the 
timing trials graphs that you have found.


bengt