Advanced search

Forums : Wish list : Initial Replication
Message board moderation

To post messages, you must log in.

AuthorMessage
Desti

Send message
Joined: 9 Jul 07
Posts: 12
Credit: 279,092
RAC: 0
Message 911 - Posted: 10 Jul 2007, 19:48:17 UTC

Hello,

I've seen that Cosmology uses an initial replication of 4 workunits. I think this is a big waste of computing resources, because the quorum for a valid result is only 2.

Bringing down the initial replication to 2 workunits would nearly double your throughput, only failed results need a third or fourth chance.

Linux Users Everywhere @ BOINC
ID: 911 · Report as offensive     Reply Quote
Profile Scott
Volunteer moderator
Project administrator
Project developer
Avatar

Send message
Joined: 1 Apr 07
Posts: 662
Credit: 13,742
RAC: 0
Message 912 - Posted: 10 Jul 2007, 20:45:06 UTC - in response to Message 911.  

Hello,

I've seen that Cosmology uses an initial replication of 4 workunits. I think this is a big waste of computing resources, because the quorum for a valid result is only 2.

Bringing down the initial replication to 2 workunits would nearly double your throughput, only failed results need a third or fourth chance.

OK, we'll play your wily game. I've changed it for now; let's see what happens.
Scott Kruger
Project Administrator, Cosmology@Home
ID: 912 · Report as offensive     Reply Quote
Desti

Send message
Joined: 9 Jul 07
Posts: 12
Credit: 279,092
RAC: 0
Message 914 - Posted: 10 Jul 2007, 20:52:19 UTC - in response to Message 912.  

Hello,

I've seen that Cosmology uses an initial replication of 4 workunits. I think this is a big waste of computing resources, because the quorum for a valid result is only 2.

Bringing down the initial replication to 2 workunits would nearly double your throughput, only failed results need a third or fourth chance.

OK, we'll play your wily game. I've changed it for now; let's see what happens.



WOW, that was a very fast response! :-)

Linux Users Everywhere @ BOINC
ID: 914 · Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 25 Jun 07
Posts: 77
Credit: 4,853,984
RAC: 0
Message 917 - Posted: 10 Jul 2007, 22:21:39 UTC
Last modified: 10 Jul 2007, 22:22:08 UTC

I agree, IR > quorum is a waste of cruncher's time. However, I also understand that some projects need to get quorums quickly, and don't want to wait additional time if someone doesn't return a result.

An alternative solution would be to implement server-side abortion. SETI does this now. For example, the project could set IR at 4 and quorum of 2. Once they have met the quorum, and remaining results that have not yet been started, are automatically aborted from the clients. Any results already in progress will continue, and still get credits upon validation.
Dublin, CA
Team SETI.USA
ID: 917 · Report as offensive     Reply Quote

Forums : Wish list : Initial Replication