Re: TFM + RSA + Dual Core :-)

From: Felix Rawlings (fraw_at_yahoo.com)
Date: 08/06/05


Date: Sat, 06 Aug 2005 18:34:18 GMT

On Sat, 06 Aug 2005 07:54:27 -0700, Tom St Denis wrote:

> Felix Rawlings wrote:
>> That stuff is old, and from Intel. The claim was from HP, and the numbers
>> quoted imply some 257000 CPU cycles for the 1024-bit private key
>> decryption operation, on a single core CPU. Like I said, it was posted
>> here a couple of years ago.
>
> Care to provide a link? Googling the web and sci.crypt has NOT turned
> up anything that contradicts what I've said so far.

        Like I said, it was somebody posting here.

> I don't see what the point your posts serve. I'm just presenting what
> a dual-threaded approach can do. If you're claiming that [effectively]
> custom logic can do better well then "duh" is my response.

        I am merely pointing out that there are reports concerning the
performance delivered by Itanium 2 in this area. In that light, it is
worth the while noticing that software running on a single core Itanium 2
processor delivers roughly the same performance as dual core Athlon
running at a CPU frequency over twice as fast as Itanium 2's. That's all I
am saying.
 
        Quite frankly, I am surprised at your touchy reaction. I am just
illustrating what kind of performance each of these two particular
architectures has so far been able to deliver on this particular problem.