Re: [Chicken-users] Nursery sizing considered stupid

[Brandon J. Van Every]

felix winkelmann wrote:
> On 7/21/06, John Cowan <address@hidden> wrote:
> > My proposal is the same as it was in my original message, and is simple.
> > Forget the variable-size nursery altogether and go with 128K, period.
> > Make this tweakable in the configuration file, and for anything else,
> > people can use -:s.  You're already halfway there.
>
> John:
>
> cd to benchmarks, compile one (say "csc -Ob dynamic"), run it with
> nursery settings -:s32k, -:s300k and the default, multiple times.
> Then tell me the nursery size doesn't make a difference. It does, at 
> least
> on Linux machines.


Rigor requires that you each specify your test machines.  "It makes a 
difference on Linux" is just as meaningless in the abstract as "it makes 
no difference, go with 128K."  I'm prepared to believe that the nursery 
size *can* make a difference on some machine, with some OS, and some 
compiler, with some optimization settings.  I do not, at present, 
believe that nsample is a rigorous or particularly reproducible test.  
The explosion of variables that can go into a rigorous test, is why I'd 
like to offload this problem to someone who thinks it's more kewl.  I'm 
going to worry about performance benchmarking *after* I've got 3D shapes 
spinning around.


Cheers,
Brandon Van Every