[Shootout-list] Directions of various benchmarks
Bengt Kleberg
bengt.kleberg@ericsson.com
Fri, 27 May 2005 16:40:36 +0200
On 2005-05-27 15:24, John Skaller wrote:
...deleted
> Um .. I need to be able to run the tests and have them
> finish in an 'estimable' and 'controllable' time, so my
> computer is useful for things other than running tests.
>
> Exactly how to control that -- for either my tests or the
> Shootout -- is a difficult issue I think.
if something is difficult to get exactly right it is sometimes ok to
start with almost right, and see if that is sufficient. in this case it
would be possible to assume that we have x seconds (3600?) and divide
with tests*languages => maxtime seconds (or maybe maxtime*2). not very
exact, but perhaps ok?
...deleted
>>do you mean result as in the current 3 metrics (time, memory-usage and
>>loc) or result as in the result of a test?
>
>
> Good question.
i propose that the result is not the metrics (just to get names sorted
out). the metrics are always of interest for a successfull test. the
result is only of interest once per test (ie, the result is the same for
all languages).
if the test fails, then the result is of interest (to help with finding
the error). no metric is of interest.
...deleted
>
> The initial design wouldn't permit that, because it doesn't make
> sense to compare 'gcc' on one test with 'gcc -O3' on another.
i agree, but the current shootout does not. so if this design is
supposed to lead us towards a better shootout this proposal should not
make too much sens (supposed to be slightly funny).
...deleted
>
> Good point: you're saying the existing minibench already produces
> enough data, it just needs to be collated.
i was not trying to say that. it could be true, none the less.
i weas trying to mention that the shootout has many uses, but the main
use is to compare all languages at their best. so that should be the
simplest data set to get hold of.
>>one thing i consider important is to have a per language based
>>configuration. i would appreciate if i could add flags for a new
>>language without editing a file with lots of other language flags in it.
>
>
> Do you mean 'translator' rather than 'language'?
there are cases where the run time is influenced by flags passed to the
byte code virtual machine. when you say translator i think of the
compiler that produced the byte code.
this could be a problem on my part of understanding what names mean.
bengt