Interesting. Do I read that report correctly if it says that GMP-4.0 is 2.5 times faster than openssl on RSA (6.3 ms for GMP, 16.3 for openssl) on the test machines? I know Torbjörn has put a *lot* of effort into GMP optimizations, for both x86-variants and a bunch of other cpu:s, but that's a huge difference.
Also, if you want to get the best performance out of GMP, you should always compile it on the machine where you will run it (or be very careful with the configure options), so that it gets the optimizations and assembler code that is tuned to that particular cpu type.
/ Niels Möller ()
Previous text:
2003-01-28 19:43: Subject: Re: OpenSSL wrapper vs Pike's SSL (Was: Bz2)
Pär Svensson drew the same conclusion (that upgrading GMP would be good) in this thesis work:
http://www.ep.liu.se/exjobb/isy/2002/3308/ http://www.ep.liu.se/exjobb/isy/2002/3308/exjobb.pdf
/ Johan Schön (Firefruit)