Re: 5.1 a little slower than 5.0.2

[Ben Sunshine-Hill <sneftel@...>]

seems a kind of silly benchmark; the bulk of the code does absolutely
nothing with garbage collected types, which is the major area where
5.1 does better. When I need to do math ops huge arrays, I do it on
the C side anyway. This is why "shootout" style benchmarks aren't
usually very useful: the type of programming they encourage is not the
same as the type of programming that usually goes on in production
code.

Ben


On Fri, 24 Sep 2004 18:01:39 -0500, Tom Spilman <tom@sickheadgames.com> wrote:
> 
> FYI.  This is a comparison of the array test from the language shootout run
> on my P4 2.8GHz with 1GB of memory.  The code:
> 
> function array_test( n )
> 
>        local x, y = {}, {}
> 
>        for i=1,n do
>                x[i] = i
>                y[i] = 0
>        end
> 
>        for k=1,1000 do
>                for j=n,1,-1 do
>                        y[j] = y[j] + x[j]
>                end
>        end
> 
>        return y[1] .. " " .. y[n]
> end
> 
> I wrapped it with os.clock() calls and got these results:
> 
> Lua 5.0.2
>  1000 - 0.312s
>  3000 - 0.891s
>  5000 - 1.515s
>  7000 - 2.11s
> 
> Lua 5.1 (work2)
>  1000 - 0.343s
>  3000 - 1.032s
>  5000 - 1.718s
>  7000 - 2.407s
> 
> This is as compiled on VC++ .NET 2003.  Looking at my old tests it seems
> like 5.0.2 compiled with VC++ 6 is faster at less than .3s for n=1000.  I
> need to reinstall 6 and check this again to be sure, but has anyone else
> noticed this?
> 
>  Tom Spilman
>  Co-owner | Programmer
>  www.sickheadgames.com
> 
>