Re: [Gnu3dkit-dev] Ann: GeometryKit 0.7.5

[Philippe C . D . Robert]

On Tuesday, October 15, 2002, at 04:39  Uhr, Brent Gulanowski wrote:
> Another thing, but this is more a question than anything else. 
> Regarding:
>
>   tuple[0] /= aScalar;
>   tuple[1] /= aScalar;
>   tuple[2] /= aScalar;
>
> Does the compiler automatically optimize this to:
>
> float temp = 1/aScalar;
> tuple[0] *= aScalar;
> tuple[1]*= aScalar;
> tuple[2]*= aScalar;
>
> or am I behind the times in thinking that division is a lot slower 
> than multiplication (meaning that my "optimization" is really a waste 
> of effort)?

I don't know if gcc does it, you can verify it easily though (use -s 
when compiling). Anyway, I often use this tmp/mulu approach elsewhere. 
So if you send me a diff or point me to the place where I forgot to do 
so, I can change it in CVS (or do it yourself, you should have access).

Thanks,

-Phil
--
Philippe C.D. Robert
http://www.nice.ch/~phip