Hi Niels,
> Not sure what would be the appropriate way to benchmark; could look at a
> range of unbalancedness ratio, at one or a few fixed sizes relevant to
> toom22 and toom32, or look at fixed ratio, perhaps 1/2, over a range of
> sizes.
I suggest taking fixed non-rational ratios, for e
Niels Möller writes:
> Bottom line: If mpn_mul is fixed to not call toom32 in the above range
> (toom22 could be a reasonable replacement), and stricter ranges for
> toom22 and toom32 are properly documented, then we can change toom32 to
> not recurse using the top-level mpn_mul. And it should al
herm...@stamm-wilbrandt.de writes:
> I would have awaited different runtimes for dferent exponents, but
> obviously different values of b do the 2803 squarings (mod p) in
> different runtimes [s]:
> min/avg/max = 573.6/584.6/590.6
>
> It is not only the difference in runtimes, but also the distrib
