memory problem with GMP 5

[Paul Zimmermann]

       Torbjörn,

> have you checked balanced multiplies too?  Do they, too, require much
> more memory than 4.3?

yes. With the same program and sa=sb=31225173 (same product size), GMP 4.3.2
uses 1.6g (same memory) and GMP 5.0.0 uses 2.1g (slightly less than with
unbalanced operands, but still more than 4.3.2).

Paul