Micro-GMP

[paul zimmermann]

       Dear Marco,

> mpz_cmpabs_ui (const mpz_t u, unsigned long v)
> {
>   int ulongsize = GMP_ULONG_BITS / GMP_LIMB_BITS;
>   mp_limb_t ulongrem = 0;
>   mp_size_t un = GMP_ABS (u->_mp_size);
> 
>   if (GMP_ULONG_BITS % GMP_LIMB_BITS != 0)
>     ulongrem = (ULONG_MAX >> GMP_LIMB_BITS * ulongsize) + 1;

is GMP_ULONG_BITS % GMP_LIMB_BITS != 0 allowed in GMP and/or asl.h?

>   if (un > ulongsize && (u->_mp_d[ulongsize] >= ulongrem || un > ulongsize
> + 1))
>     return 1;
>   else
>     {
>       unsigned long uu = mpz_get_ui (u);
>       return GMP_CMP(uu, v);
>     }
> }
> 
> 
> When mp_limb_t is unsigned long, gcc -O2 compiles this code exactly as
>   if (GMP_ABS (u->_mp_size) > 1)
>     return 1;
>   else
>     GMP_CMP(mpz_get_ui (u), v)
> [...]

sounds good.

> The same as the current code. And it gets optimised also for uint16 or uint8.
> 
> >> You also changed mpn_invert_3by2. Maybe a handful of (mp_limb_t) casts
> >> could be added to the original function to make it work... but your
> >> rewrite is simpler and faster. Maybe the same could be done on
> >> gmp_umul_ppmm? for speed?
> >
> > indeed, in several places I opted for simplicity, since speed was not
> > (at first) a goal. But when running "make check" of MPFR with micro-gmp-8,
> > I now realize speed is also important!
> 
> A attach a possible implementation of mini-gmp that should support limbs
> of "any" size. Can you test it with MPFR? Both correctness and speed...

I'll do.

Paul