mpf exponent range

Rick C. Hodgin foxmuldrster at
Tue Aug 16 18:02:37 CEST 2011


My opinion is always to do the utmost in service to others.  While it may
not be something the overwhelming vast majority of people need, some people
might need it.  It is for that group that your work should be directed, so
that when there is a need, by that person or another, it will be available.

My philosophy for developers is this:  If something is possible and may have
utility for those few pushing the absolute extreme frontiers of needs or
wants in computing, why not do it?

My opinion.

- Rick

-----Original Message-----
From: gmp-discuss-bounces at [mailto:gmp-discuss-bounces at]
On Behalf Of Torbjorn Granlund
Sent: Tuesday, August 16, 2011 10:50 AM
To: gmp-discuss at
Subject: mpf exponent range

We recently improved the logarithm calculations needed for input and output
conversion of mpf numbers.  Before this fix, things started to work
unreliably for exponents (in some base) of over about 2^50.

Now, input and output work for numbers up to about 2^(2^63-1) (for 64-bit
machines).  During computations, much larger numbers, up to
(2^64)^(2^63-1) are expected to work.

I realise mpf isn't terribly much used today, given the existence of the
superior mpfr functions.  Do people on this list have any opinions about the
exponent range, of mpf, in particular, if we should make input and output
work for the entire allowed number range?  Or is this irrelevant?

It is not a trivial fix.  The required changes will be to allow multi-word
exponents (for bases < 2^64).

gmp-discuss mailing list
gmp-discuss at

More information about the gmp-discuss mailing list