Why do I never get more then half of my requested credit?

Message boards : Number crunching : Why do I never get more then half of my requested credit?

To post messages, you must log in.

AuthorMessage
nick n
Avatar

Send message
Joined: 26 Aug 07
Posts: 49
Credit: 219,102
RAC: 0
Message 57658 - Posted: 6 Dec 2008, 18:48:53 UTC

I have never been able to get more then 1/2 of the requested credit on my iBook. Why is that and can it be fixed? This is the computer I am talking about https://boinc.bakerlab.org/rosetta/results.php?hostid=936171 Thanks!
ID: 57658 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
mikey
Avatar

Send message
Joined: 5 Jan 06
Posts: 1895
Credit: 9,179,826
RAC: 3,209
Message 57735 - Posted: 9 Dec 2008, 11:21:45 UTC - in response to Message 57658.  
Last modified: 9 Dec 2008, 11:24:46 UTC

I have never been able to get more then 1/2 of the requested credit on my iBook. Why is that and can it be fixed? This is the computer I am talking about https://boinc.bakerlab.org/rosetta/results.php?hostid=936171 Thanks!


Honestly it looks like it is happening with all your machines, not just the one. It looks like it is happening to everyone, not just you. Click on anyone elses name in the left hand column, then view their computers, then click on the tasks and you will see that almost all requested credits are around twice the granted credits. Not everyone lets you view their computers, but enough do to let you see what I mean.
ID: 57735 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 57748 - Posted: 9 Dec 2008, 14:31:46 UTC

The claims are actually rolled forward into the moving average. That is what granted credit is based upon, average credit claim, across the whole project for that protein, per model. So, *somewhere* in the project you will also find a similar number of TFLOPs crunching that are getting more then their claim (although perhaps just 10% more on a larger number of machines). It's just how the process works.

When you find the machines getting more then their claim it will always raise the question of bias, then a lengthy discussion of benchmarks will ensue. Bottom line is that running Rosetta for that specific protein is a benchmark as well. And some machines and architectures do better at it then others. With another protein and a different mixture of calculations and memory accesses, the tide may turn towards some other environment doing better. The application is not coded nor optimized to have any preferential bias.

In general, L2 cache is very important. And, just recently, the tasks have been using significantly more memory then previously. So making sure BOINC can access enough memory to run the work efficiently is important as well.
Rosetta Moderator: Mod.Sense
ID: 57748 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dcdc

Send message
Joined: 3 Nov 05
Posts: 1832
Credit: 119,675,695
RAC: 11,002
Message 57749 - Posted: 9 Dec 2008, 14:40:59 UTC - in response to Message 57735.  

I have never been able to get more then 1/2 of the requested credit on my iBook. Why is that and can it be fixed? This is the computer I am talking about https://boinc.bakerlab.org/rosetta/results.php?hostid=936171 Thanks!


Honestly it looks like it is happening with all your machines, not just the one. It looks like it is happening to everyone, not just you. Click on anyone elses name in the left hand column, then view their computers, then click on the tasks and you will see that almost all requested credits are around twice the granted credits. Not everyone lets you view their computers, but enough do to let you see what I mean.

have a look at pentium D machines - they tend to get more than requested, i assume because their benchmarks are relatively low.
ID: 57749 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Why do I never get more then half of my requested credit?



©2024 University of Washington
https://www.bakerlab.org