> Bodo> Quite a few of these files are just the assember output from gcc since on
> Bodo> quite a few machines they are 2 times faster than the system compiler.
>
> gcc hasn't produced better code that Sun's cc for SPARC in eternities.
Not if you tell gcc to generate v8 (meaning hardware multiplication),
but not to Sun's cc. And it was like this as far as I can remember back
to SSLeay days...
> If we
> want to distribute fast crypto routines, we should either hand-create them,
Why? I mean "distribution" isn't the reason, is it? The reason is that
we want them to *be* fast, isn't it? :-) In either case my judgment is
that only bn_*_comba[48] and bn_div_words are worth the effort (provided
that we're talking about pure v8). I don't beleive you'll be able to
squeeze more than 5% out of the rest if you hand-code 'em. As for v9
(and some extra thoughts on v8) see my previous post on the matter...
Cheers. Andy.
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
