[RC5] Client for Windows/64Bit

Fuzzy Logic fuzzymo at gmail.com
Mon Aug 15 21:38:02 EDT 2005


I do crypto work with Linux on both 31-bit and 64-bit platforms (mainframes).

I've never seen a 64-bit application work faster than a 31-bit
application on the same hardware. Never.

Of course, none of the work I've done recently uses hand-coded
assembler routines for crypto operations, so it is possible to improve
the speed IF the operation can take advantage of larger registers,
which is certainly true for DES operations, but I'm not familiar
enough with the OGR algorithm to know if 64-bit registers would help.

Fuzzy

On 8/15/05, Scott Dodson <sdodson at sdodson.com> wrote:

> Ya, while there's a x-64 version of XP available very few are running it.
> 
> This past weekend I had the pleasure of being part of a team assisting
> about 4,000 students move into dorms and get connected to the Internet.
> Surpsingly there were a great deal of machines branded AMD64, and
> several Intel EMT64 machines though those were harder to spot.  Yet none
> of the machines I encountered actually had 64bit versions of XP on
> them.  I was a bit surprised by this, but oh well.  XP-64 was released a
> few months ago and we're just now seeing mainstream driver support for
> it, I imagine OEMs will switch soon.  I think we'll see the fastest move
> to 64bit when Vista is released.  It's taken about two years for the
> chip to become mainstream, I imagine it will take another year or so for
> software vendors to catch up.
> 
> Besides, we'd need someone really good at amd64 optimization before
> you'd see any dramatic improvements.
> 
> Fuzzy Logic wrote:
> 
> > I'm not so sure of that. It is possible that the 32-bit client would
> > perform better than a 64-bit client. Also, currently, the gain for
> > creating a 64-bit client would be small given the small number of
> > people using a 64-bit processor AND a 64-bit version of Windows (which
> > is still in beta test anyway).


More information about the rc5 mailing list