OT Linux (RE: [rescue] OT: Stuffed Proliant?)

[Patrick Giagnocavo]

On Sat, Dec 22, 2001 at 01:55:12AM -0500, Joshua D Boyd wrote:
> On Sat, Dec 22, 2001 at 01:09:01AM -0500, Patrick Giagnocavo wrote:
> > On Sat, Dec 22, 2001 at 12:30:27AM -0500, Joshua D Boyd wrote:
> > > On Sat, Dec 22, 2001 at 12:01:56AM -0500, Patrick Giagnocavo wrote:
> > > Yes, but how hard do you really push the K6-2?
> > 
> >I don't push it at all... my customers who run db-backed web servers do though
> > 
> > One db query at least for each page load.
> 
> Yes, but how many hits per second, how tough of a query per page, how large the
> data set, run times involved per hit.
> 
> The one thing that intel does have going for it is how cheaply you can bring a 
> second machine online (assuming you designed your site correctly).  This can
> be used to overcome a lot of short comings.

Actually, you could probably get better performance from other
platforms, but it is easier to solve problems with more RAM, which is
cheap on x86.  DBMS's luuuuuve more RAM.
 
> I don't know.  It would interesting to see for common sites how much of the 
> load is the SSL processor versus everything else.  I'm thinking that based
> on much CPU power is used on my machines for SSH that it isn't much.  Err, both
> SSH and SSL use RSA, do they not.

All I know is that many people were very happy that the crypto add-in
cards got supported in OpenBSD and Linux, etc.  Could also be for
IPSEC/VPN usage though.

> Besides, lack of rotate isn't all that bad.  Instead of rotating left by 12 
> bits, you shift left by 12, shift right by 20, then sum the two shifts.  The
> two shifts can be done simultaneously on a super-pipelined processor, so you
> need 3 cycles to do a rotate instead of one.  While this seems bad, I find it

As long as the cache stays full...

> hard to believe that this is the sole reason for distributed net to be slower
> on SPARCs than intel boxes.  We loose 2 cylces on a rotate.  So what.  We
> make it up by doing less loads.

Maybe loads aren't as expensive when everything is in L2 cache.

./patrick