GMP5 benchmark

[James Wanless]

On both GMP5.0.0 and GMP5.0.1 I experienced similar behaviour (IIRC):
Because I'm using a slightly old version of Xcode (3.1.2) on Mac Intel  
Leopard (10.5.8), I needed to specifically tell GMP to use the 4.2  
version of gcc with:
"./configure 'CC=gcc-4.2' 'CXX=g++-4.2'"
J

On 16 Feb 2010, at 07:36, Torbjorn Granlund wrote:

> James Wanless <james at grok.ltd.uk> writes:
>
>  Good speedup w/ GMP5 vs. GMP4.3.1:
>  One of my programs, factorize6, running on (M+2)1257787, has  
> execution
>  time reduced from ~4358 mins to ~1716 mins!
>
> Nice, that's more than one could usually hope for with 5.0.1.
>
>  thanks v. much to the GMP developers (and to Richard Dickerson for a
>  suggestion on gmp-bugs that enabled me to build on Leopard),
>
> Did you have build problems on Mac OS "Leopard" with GMP 5.0.1, or are
> you talking about 5.0.0?
>
> (Note that 5,0.1 s actually faster than 5.0.0, since we provided  
> better
> tuning tables, and fixed some memory greed problms.)
>
> -- 
> Torbjörn