RE: [TCLUG:3846] rc5 speed on linux

["Eric Hillman" <ehillman@cccu.com>]

> -----Original Message-----
> From: Christopher Palmer [mailto:reid@pconline.com]
> Sent: Tuesday, February 02, 1999 1:31 PM
> To: tclug-list@listserv.real-time.com
> Subject: Re: [TCLUG:3846] rc5 speed on linux
>
>
> On Tue, 2 Feb 1999, Ben Luey wrote:
>
> > Or are k6's just bad at this type of calculation?
>
> I think I recall something on tomshardware.com about how the clones aren't
> as good at floating point as the true Intel x86 chips. But then, I don't
> know if cracking keys is a floating point operation or not.
>

	It's all integer math (so far as I know), but the newer rc5 clients make
use of the Intel MMX instruction sets, giving them a serious edge as far as
rc5 goes.
	The folks at distributed.net'll tell you though, that keycracking speed
isn't really a good measure of processor performance, though.  RISC chips
tend to suffer, for example, even though RISC chips are, so they tell me,
far more efficient at most computing tasks than x86's.