[Shootout-list] fannkuch (timer resolution; HZ=1000?)

[Bengt Kleberg]

On 2005-05-24 19:33, John Skaller wrote:
...deleted
> 
> The 'best way(tm)' is to include the startup time in the result.
> For low n, this will punish some systems (eg Java). For high
> n it will reward some systems (eg Java) when the startup costs
> bear fruit (eg Java JIT operation).

afaik it is possible with the current shootout to include the startup 
time in the result.

i think that the reson to remove it (for those that want to use the 
possiblity) is that the test have a finite amount of time (24 hours) in 
which to be run. having long running tests makes it impossible to get 
all the tests and all the languages into that total time frame. so 
shorter tests have to be used. and shorter tests suffer severly from 
start up costs. thus there is a way to remove that cost, for those so 
inclined.
given the 24 hour limit it sounds like the best way to me.


> The real problem is that CRAPS concentrates on the times for
> high n -- the hello test being the exception. Perhaps it should

i do not think craps is interesting so i do not think it is a problem. 
except when it is used as a reason to not measure more data points for 
the graphs. then it is a problem.


bengt