On 9/18/05, ben(_at_)algroup(_dot_)co(_dot_)uk (Ben Laurie) wrote:
Ian G wrote:
Numbers?
I was going to provide them, but accurate numbers for DSA is difficult,
because it requires modifying the core code in OpenSSL (or some other
DSA generator), and I couldn't be bothered.
One can get a feel for it with:
time openssl gendh <number of bits>
I'm not sure, based on these numbers, that this benchmark procedure is
reproducible enough to give valid data with only one sample per prime size.
However, here is what I got.
On a Mac G4/450MHz running OS 10.3.9:
[G4:~] billfran% time openssl gendh 512
...
9.850u 0.170s 0:18.39 54.4% 0+0k 3+7io 0pf+0w
[G4:~] billfran% time openssl gendh 1024
...
11.920u 0.070s 0:13.19 90.9% 0+0k 0+3io 0pf+0w
[G4:~] billfran% time openssl gendh 2048
...
5909.970u 24.060s 1:45:08.60 94.0% 0+0k 0+4io 0pf+0w
[G4:~] billfran% time openssl gendh 3072
...
21017.450u 166.540s 6:42:28.96 87.7% 0+0k 0+3io 0pf+0w
Cheers - Bill
-----------------------------------------------------------------------
Bill Frantz | gets() remains as a monument | Periwinkle
(408)356-8506 | to C's continuing support of | 16345 Englewood Ave
www.pwpconsult.com | buffer overruns. | Los Gatos, CA 95032