Subject: Re: AXPpci 33 ... not very fast?
To: Curt Sampson <cjs@portal.ca>
From: Eric Sentner <sentner@katana.digex.net>
List: port-alpha
Date: 04/25/1997 16:31:54
References: <199704252024.QAA27103@katana.digex.net>
	<Pine.NEB.3.96.970425132657.2463a-100000@cynic.portal.ca>


On April 25, Curt Sampson wrote:

> On Fri, 25 Apr 1997, Eric Sentner wrote:
> 
> > Doing an ssh-keygen on my 233Mhz Multia
> > took about 5 seconds compared to the 15 seconds it took on my sparc5
> > desktop machine at work, and the 3 minutes it took on my 486/50.. :)
> 
> Actually, I have noticed that my 166 MHz AXPpci33 generates its
> session key in about half the time my P166+ does, and ssh logins
> to that machine are noticably faster than to any other machine I
> own.

Yeah, the integer math on these things kicks butt.. :)


> On the other hand, I spend a heck of a lot of time compiling....  :-)

We (Some of the systems engineers where I work) had a big discussion a
few months ago about this.. We came to the conclusion that it is slow
for compiling because it's RISC and has align the code to 64-bit
boundaries which takes a bit more processor time to optimize it.. But
once the stuff is compiled it's generally faster.. Floating point is
probably faster on the Pentium but not integer.. :-)

-- Eric