Bald andere Applikation und anderes Creditsystem

gruenmuckel

Grand Admiral Special
Mitglied seit
17.05.2001
Beiträge
29.626
Renomée
1.743
Standort
Gerry-Weber - Stadt
  • SIMAP Race
  • QMC Race
  • RCN Russia
  • Spinhenge ESL
  • BOINC Pentathlon 2012
  • BOINC Pentathlon 2014
  • BOINC Pentathlon 2016
  • BOINC Pentathlon 2018
  • BOINC Pentathlon 2020
  • BOINC Pentathlon 2021
Laut der aktuellen Newsmeldung dort sieht es so aus das in etwa 2 Wochen die hmmer Application rauskommt und gelichzeitig fixed Credits per WU.

Also werden wir dann keinen Vorteil mehr daraus ziehen das bei uns praktisch jeder hohe Credits claimt.

Deswegen schlage ich vor solange wir diesen kleinen Vorteil noch genießen, ihn zu nutzen.
Also bis zur neuen App rechnen was geht.
http://boinc.bio.wzw.tum.de/boincsimap/forum/viewtopic.php?t=446




Bird-dog



Joined: 27 Apr 2006
Posts: 2
Location: UK

PostPosted: Thu Aug 31, 2006 12:20 am Post subject: New quorum and credits Reply with quote
This from the homepage

August 30, 2006
Lower initial replication and quorum: With the release of the SIMAP applications 5.10 and 5.11 the error results have been drastically reduced. We are very happy about this and would like to thank all participants for the recently increased power of our project (more than 5 TLOPS now!). It is now safe for the scientific parts of BOINCSIMAP to switch back to a minimal quorum of 2 and an initial replication of 2 as we used when BOINCSIMAP started. Recently we have decreased the initial replication to 3. To address the inhomogeneously claimed credits we will switch to fixed credits per workunit that will be determined by the number of sequences in the workunit and the number of hits in the result. We will test this system first with the second application hmmer which is about to start and then apply this to the simap app too (and decrease the quorum to 2). Please use our message boards to discuss the fixed credits per WU. We are interested in your opinion about this.
Back to top
View user's profile Send private message
kevint



Joined: 15 Dec 2005
Posts: 7


PostPosted: Thu Aug 31, 2006 2:37 am Post subject: Reply with quote
I think this is a step in the correct direction, It will be interesting to see how this will affect the credits.

With optimized BOINC client this might reduce the amount of credits per WU, while increasing credits per WU for those not using the optimized BOINC apps.

It should also even out the differences I have seen between AMD and Intel based processors. AMD has always done better on this project than Intel, and since most of my crunchers are Intel based it will be interesting to see what happens.

Any idea when this will be released into production ?
Back to top
View user's profile Send private message
rattei
Chef


Joined: 09 Nov 2005
Posts: 187


PostPosted: Thu Aug 31, 2006 7:39 am Post subject: Reply with quote
kevint wrote:
Any idea when this will be released into production ?


Hopefully in 2 weeks. Our schedule is to launch the hmmer app first, see how the fixed credits work for it and apply them then to the simap app.

Best regards
Thomas
 
Obwohl ich die optimierten Clients nutze finde ich das System mit festen Credits natürlich viiiiel besser! Daumen hoch dafür
 
dazu auch

September 1, 2006
hmmer tests starting: Currently we are running tests with our new hmmer application based on a few workunits. So be not surprised if the hmmer app appears on your workstation. Large scale alpha tests will follow soon.

und was ist hmmer

The hmmer application is an additional application which extends the SIMAP contents. The simap application compares proteins by their similarity. The results are stored in the SIMAP database and are used by scientists using Web pages, Web services or locally installed programs on our servers.

The hmmer application extends this concept to "protein domains". Those domains are mathematical models that describe the functionally relevant and unique parts of proteins. The type of model we are using is called "Hidden Markov Model" (short HMM). The HMMs are the most powerful technique to mathematically describe protein domains and store them into databases. Using the hmmer application we want to compare all known proteins with all protein domain models in several domain databases, in order to find functionalities of proteins. This data is very useful to predict the functions of the proteins - a result that many biologists are extremely interested in.

http://boinc.bio.wzw.tum.de/boincsi...419&sid=237749b5213f44b438ce6fcf6df83aba#4419

mfg
Sir Ulli
 
Zurück
Oben Unten