Re: pidigits benchmark

[Isaac Gouy <igouy2@...>]

--- Mark Meijer <meijer78@gmail.com> wrote:

> > Except when it is the same as benchmarking those programming
> languages
> > themselves - when what appears to be an implementation in the
> > programming language just gets transformed into bindings to a
> native
> > library - is it turtles all the way down?
> 
> And I'm unimaginative ;-) 

I hope you don't think it's nit-picking if I remind you that I said the
kind of complaint you made was unimaginative - afaik you may be the
Leonardo Da Vinci of the age ;-)


> Sure you can argue that everything is
> ultimately native, since everything ultimately translates to
> instructions being executed on your CPU.

Not what I meant - I've been told by the implementer that the
apparently "pure" arbitrary precision arithmetic in one language
implementation really is just a sequence of GMP calls. 


> Granted, it can be hard to
> draw the line (or impossible, processors exist that execute java
> bytecode in hardware, for example). But I still think running
> benchmarks of algorithms written completely in a native compiled
> library, tells me more about that library than the language used to
> invoke it. 

If the library is a constant factor, why would this tell us more about
the library than the (variable factor) language used to invoke it?


-snip-
> > I like to see both ;-)
> 
> Then you agree the distinction I made is a valid one.

iirc my objection was to "what is the point" and "kinda silly".