Message boards : Number crunching : not sending workunits to multiple users
Author | Message |
---|---|
Dragokatzov Send message Joined: 5 Oct 05 Posts: 25 Credit: 2,446,376 RAC: 0 |
I am confused by this. why is rosetta not sending out the same workunit to 3 or 4 people to make sure its crunched? Victory is the ONLY option! |
Kerwin Send message Joined: 19 Sep 05 Posts: 10 Credit: 1,773,393 RAC: 0 |
Hi Dragokatzov, the short answer is that Rosetta doesn't rely on a quorum of 3 or 4 results before granting credit, unlike the other BOINC projects. What you claim is what you get. |
Marky-UK Send message Joined: 1 Nov 05 Posts: 73 Credit: 1,689,495 RAC: 0 |
Also, if a WU fails or times out, it gets resent to another user, so someone else can have a crack at it. (If it fails a second time it doesn't get resent again.) This approach maximises the number of WUs that get crunched. It also creates a different problem (credit) but that's another story... |
Dimitris Hatzopoulos Send message Joined: 5 Jan 06 Posts: 336 Credit: 80,939 RAC: 0 |
In thread Validation it is explained why Rosetta runs with "initial replication"=1, quorum=1 Basically, when testing a protein, we all get the same raw protein (aminoacid) data (same WU), the difference being the random seed (i.e. the initial instructions to our "virtual explorer" looking for the lowest elevation point). Best UFO Resources Wikipedia R@h How-To: Join Distributed Computing projects that benefit humanity |
tralala Send message Joined: 8 Apr 06 Posts: 376 Credit: 581,806 RAC: 0 |
I am confused by this. why is rosetta not sending out the same workunit to 3 or 4 people to make sure its crunched? Another way to explain is it that they need only the "best" (correct) result. Verifying the best result is easy and can be done independently. All other runs are actually tries to find the "best" (correct) result and thus not necessarily all needed. |
Message boards :
Number crunching :
not sending workunits to multiple users
©2024 University of Washington
https://www.bakerlab.org