When the FFT kicks in, it is only just faster than the Toom algorithm.
So for numbers just in the FFT region you might not notice any
difference. Also, 5000 limbs was approximate. It might be 7000. It
depends on your architecture.

As far as we know MPIR 0.9.0 is highly stable and correct, so you
shouldn't have any issues using it in production code. It is used in
Sage for example, which has thousands of users. So far there have been
no reported issues (except build issues on strange hardware, which you
obviously haven't had).

Bill.

2009/2/27 Jeff Gilchrist <jeff.gilchr...@gmail.com>:
>
> On Thu, Feb 26, 2009 at 5:15 PM, Bill Hart <goodwillh...@googlemail.com> 
> wrote:
>
>> The FFT currently kicks in at about 5000-7000 limbs or so. Given that
>> each 64 bit limb is about 19 decimal digits that is at least 100,000
>> decimal digits before the FFT is used. In a later version of MPIR
>> we'll have an FFT which kicks in much earlier (about 2000 limbs), but
>> it is a big project to change it over.
>
> That would explain why the different FFT tuning parameters didn't make
> a difference if the FFT code doesn't start until somewhere around
> 100,000 digits.  I'm not running something around 105,000 digits so
> hopefully that will be using the FFT code.
>
>> I guess you know that tuneup only prints the values out. It doesn't
>> change the values in the relevant gmp-param.h.
>
> Yes, I update gmp-param.h each time and re-compile.  How would you
> suggest people tune the parameters in Windows using Visual Studio
> since there doesn't seem to be a tune application, or is there?
>
>> As for MPIR 1.0.0, we can't wait either. Soon I hope...
>
> Should I be waiting for 1.0 before I use it for production code or if
> "make check" passes I should be good to go?
>
>> Thanks for your comments and timings. That sort of feedback encourages us.
>
> My pleasure.
> - Show quoted text -
> Jeff.
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"mpir-devel" group.
To post to this group, send email to mpir-devel@googlegroups.com
To unsubscribe from this group, send email to 
mpir-devel+unsubscr...@googlegroups.com
For more options, visit this group at 
http://groups.google.com/group/mpir-devel?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to