hpf 2 5 .18

Joined
Jan 14, 2006
Messages
12,268
Reaction score
283
I'm currently Crunching uints begining in Boinc Manager hpf 2 5.18, but on results page as la277.
But it's not just 3 people cunching the unit, some have been done by 12 others all validated OK?
Why so many on 1 unit:confused:
 
This is a new way of crunching introduced for this project - they are effectively splitting each WU into 19 seperate units and sending them out, once 11 are returned they will attempt to validate.

Therefore the old quorum of 3 does not occur - there will be no-one else crunching your 'bit'.
Human Proteome Folding - Phase 2 will work differently then our other projects that we currently have running. Each computer will actually be processing different work. Please read this thread for details: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=9587

The way that this will specifically work on BOINC is that for each workunit we will send out 19 replicas of the workunit. Each result will actually compute different data and this data will be validated on the backend (again, please read the thread above for details). Validation will be attempted when the first 11 successful results are returned. Please note that while we will validate when the first 11 results are returned, we do need all the results back. Results that are returned after the quorum has been reached will still be used! We have code in place to ensure that the result was processed fully. Points will be awarded based upon the number of attempts that are made for each result. Thus when credit is computed, the system will compute the average points per attempt. This will be averaged for the results within the set. Then points will be awarded for each result by # attempts * average points per attempt.

I would encourage everyone crunching to frequently visit the WCG forums to keep up to date with the latest project info :thumb:

user.gif
 
Back
Top