Questions and Answers : Wish list : Initial Replication
Message board moderation

To post messages, you must log in.

AuthorMessage
Damaraland

Send message
Joined: 15 Jul 05
Posts: 5
Credit: 65,819
RAC: 0
Message 12092 - Posted: 16 Jan 2006, 22:50:49 UTC

Some other projects ( Einstein) has reduced it's initial replication to 3.

Wouldn't it be better to start with an initial replication of 3 units and "give the chance" of work to finish befor distributing the 5 at the beginnig?.
For example I see that right now some workunits have received 4 or 5 units, but other ones only have received 1 or 2.

I think this will improve not only Cern proyect but all BOINC proyects.
ID: 12092 · Report as offensive     Reply Quote
Profile Keck_Komputers

Send message
Joined: 1 Sep 04
Posts: 275
Credit: 2,636,117
RAC: 40
Message 12097 - Posted: 17 Jan 2006, 1:50:34 UTC

Such settings are set based on the needs of the project. Here at LHC turnaround is very important to the project. With an initial replication of 5, workunits are completed much quicker since 2 hosts must have problems before an additional result is needed.
BOINC WIKI

BOINCing since 2002/12/8
ID: 12097 · Report as offensive     Reply Quote
Damaraland

Send message
Joined: 15 Jul 05
Posts: 5
Credit: 65,819
RAC: 0
Message 12139 - Posted: 17 Jan 2006, 21:40:37 UTC - in response to Message 12097.  

Yes, but even then. There are units that are not been sent yet. Even to be fast I think it will be faster to give initial replication of 3 and when all units have work restart with the beginning.
ID: 12139 · Report as offensive     Reply Quote

Questions and Answers : Wish list : Initial Replication


©2021 CERN