Re: [flint-devel] Timing code

[William H. <ha...@ya...>]

The timings certainly sound encouraging!

I should mention that sometimes you can be misled
about a timing due to optimisations the compiler is
making. It might decide that it can rearrange things
in a different order, and then you end up timing
something you didn't intend to.

I've certainly had this happen. Usually when you are
making a call to a function in another file or module
this is not a problem, but it is definitely worth
checking by timing everything then timing just the bit
you are not interested in, and taking the difference,
to make sure the result is what you expect.

Regarding primes, Magma has special code for primes p
such that p-1 can be represented in 1, 2, 4, (possibly
also 8 and 16), 22/23, 30 or 64 bits I believe.
Strangely I've not been able to find variances at 53
bits which I found odd, since this is the size of a
double. 

I would try a small prime, e.g. p = 5, one at 22 bits,
one at 30 and one at about 62 or 63 bits. I suspect
you'll be way ahead for large primes. 

Bill.

--- Richard Howell-Peak <ric...@gm...> wrote:

> Well I'm pretty sure I don't time how long it takes
> to generate the
> polynomial anyway since I start timing just before
> and after I call
> Factorisation. I had the iterations low just so it
> would run at a reasonable
> speed on my laptop, I was thinking maybe 10 or so
> would be sufficient.
> 
> I was wondering what data ranges would be good. I
> was thinking of doing a
> few primes, a small one, a medium one and a large
> and comparing the output,
> then maybe later in the week I'll do one that runs
> through a whole load of
> primes and makes a 2x2 array for each one which I
> can make into a nice
> picture. At the moment I've only been going up to
> polynomials of length 250
> and it looks quite promising - sometimes flint is
> way ahead, sometimes magma
> has the edge and sometimes they're pretty close so
> we'll see what happens.
> 
> -- 
> Richard
> >
-------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your
> Move Developer's challenge
> Build the coolest Linux based applications with
> Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source
> event anywhere in the world
>
http://moblin-contest.org/redirect.php?banner_id=100&url=/>
_______________________________________________
> flint-devel mailing list
> fli...@li...
>
https://lists.sourceforge.net/lists/listinfo/flint-devel
>