[RC5] Proper method for importing buff-out files?

id id at sonic.net
Tue Nov 17 21:32:27 EST 1998


I think the reason we are dissapointed is because on a duel ppro 200 I got 1.2
mk/s and that system is a couple years old.  One would think that a new wiz bang
xeon would improve this per mhz. where in fact it looks a bit slower.  The ppros
have 256k cache and the xeon has 4 times that so I would expect it to be faster,
of course I don't know how much effect that really has on rc5 performance....

btw, why were you running it a half speed?

James Flom

"John \"Chris\" Wren" wrote:

>         Well, I don't know exactly why you'd find the performance of a
> Xeon disappointing.  The fact that it has a meg of *full speed* cache,
> as opposed to the 1/2 speed cache of a Pentium II makes for a major
> performance advantage.  Every statistic available shows that a 400mhz
> seriously outperforms a 450mhz PII, in all cases except where
> continous cache faults occur (and even Windows 95 doesn't cause that).
> And with the advent of the 450mhz Xeons (Tanner), a PII 400 doesn't
> come close.
>
>         And the numbers I gave were using 4 processors, not 2.  As I
> mentioned it was running at about 50% of it's rated speed (i.e., about
> 200 mhz per processor, or a little less).
>
>         While I'm aware of how much time the algorithm spends in
> cache, I'm sort of curious what the memory requirement of it is.  In
> this particluar system, the memory subsystem is running at a pathetic
> 66mhz.
>
>         When this system is brought full up, the numbers will be at
> least a factor of 2 better (perhaps even close to 3).
>
>         - Chris
>
> On Tue, 17 Nov 1998 10:58:45 -0500, you wrote:
>
> >I find the performance of the Xeon somewhat disappointing.  I expected it to
> >be a somewhat better.  I am running a PII-350 and am getting 980K kps.  What
> >is the speed of the processors?  If they are running 400Mhz that it would be
> >about right.  I expect you are lousing some performance due to the
> >multi-processor communications.  I am also assuming that it was using 2
> >processors instead of 4.
> >
> >You can stop processing on your laptop and copy your buffer files to a safe
> >location.  Copy the buffer files from the Xeon box to your RC5 directory and
> >perform a flush.  Then copy your original laptop buffers back to the RC5
> >directory and start processing where you left off.
> >
> >-John Scott
> >kd4vhg at radio.org
> >
> >
> >> ----------
> >> From:        jcwren at atlanta.com[SMTP:jcwren at atlanta.com]
> >> Reply To:    rc5 at lists.distributed.net
> >> Sent:        Tuesday, November 17, 1998 8:34 AM
> >> To:  rc5 at lists.distributed.net
> >> Subject:     [RC5] Proper method for importing buff-out files?
> >>
> >>      I recently (last night) used a 4 way Xeon box I had access to
> >> crunch keys as a demo for some folks.  Overnight it ran 336 random
> >> blocks (no network for this puppy, yet).  I'd like to submit those
> >> blocks.  I have a laptop that is connected to the network running the
> >> RC5 client.
> >>
> >>      How do I properly go about flushing the keyblocks the other
> >> machine did without disrupting the laptop?
> >>
> >>      By the way, this machine was running at about 50% it's rated
> >> speed, and was doing 2.21M KPS.  Can't wait 'til it goes faster!
> >>
> >>      - John "Chris" Wren, Team Anvil
> >>
> >> --
> >> To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
> >> rc5-digest subscribers replace rc5 with rc5-digest
> >>
>
> --
> To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
> rc5-digest subscribers replace rc5 with rc5-digest

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list