Re: Poor man's benchmarking issue - help needed

[Lorenzo Donati <lorenzodonatibz@...>]

On 31/08/2011 10.17, Alexander Gladysh wrote:
> On Wed, Aug 31, 2011 at 11:03, Lorenzo Donati
> <lorenzodonatibz@interfree.it>  wrote:
>
> > I stumbled a funny result when doing very simple benchmarking on some of my
> > code.
>
> You did not say if you were using plain Lua or LuaJIT. Seems to be plain Lua.

Yes, sorry. You are right. Plain Lua.

> > According to the following benchmarks, built-in math.floor and math.ceil are
> > slower than their counterpart rewritten using the core % operator!
>
> That is to be expected. C function call is relatively slow.

Do you mean that floor and ceil are really slower?

What's the point of including them in the library then? I assumed they 
were at least slightly optimized.

I agree that they are more readable, but since they are really easy to 
replace with a one liner using %, shouldn't they fall under the "useless 
fat" principle of Lua - If you can do them yourself easily and they 
would be even faster, why not removing them? Wasn't those the reasons 
that led to log10 removal in 5.2?

Am I missing something?


> Alexander.
thanks for the feedback.

-- Lorenzo