[RC5] stats update

Bruce Wilson bwilson at distributed.net
Wed Dec 18 09:49:59 EST 2002


| Hi,
| 
| Since RC5-72 stats are online. With RC5-64 I crunched 6000+ blocks
on 
| average a day. Now with the bigger packets (and no altivec 
| code) I only 
| do 150 packets a day ;-(

Remember that a single RC5-64 packet had anywhere from 2^28 keys to
2^33 keys.  Had you been using the maximum size, these numbers
probably would have been closer.  Don't worry, we liked the boost from
the Altivec as well - I'm sure someone is working on it.

| Since the stat-run should also take considerably less time, 
| should dnet 
| not update the stats more often. Say twice a day (as a start)...
| 
| I remember that with the RC5-56 challenge that lasted less then 24 
| hours there were live stats available.
| 
| Since more project (like d2ol) have live statistics why couldn't
dnet 
| do the same...

Realtime stats sound like a good idea, but aren't really practical in
our environment.  The ranking information is pretty meaningless if the
numbers are changing continuously.  Also, our network is designed for
reliability, not instant feedback.  It's very possible that work you
submit shortly before a stats run may not even appear in the same day
you submitted it.

We did have realtime stats during RC5-56, but quickly discovered that
instant feedback was not only problematic to deliver, but  encouraged
behavior for our network.  Participants would flush after every block,
and rapid-refresh the website to see the blocks hit.  That meant
hammering on both the database and the webserver, and we didn't like
it.


__
Bruce Wilson <bwilson at distributed.net>
PGP KeyID: 5430B995, http://www.toomuchblue.com/ 

"I want to move to Theory. Everything works in Theory."
    --John Cash, id Software


--
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