Message boards : Theory Application : New Version 263.60
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,189,247
RAC: 104,501
Message 35722 - Posted: 1 Jul 2018, 2:21:33 UTC - in response to Message 35712.  

Hopefullly Sherpa jobs are not being sent out under the multi-core plan and are being restricted to the single-core plan?

Yes, sometime they are successful. They running at the end of the possible runtime for Theory.
Magic wrote 4 Tasks with 2 CPU's are good in 8-Core PC.
This will be a good performance therefore (Sherpa).
ID: 35722 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,466,562
RAC: 103,962
Message 35724 - Posted: 1 Jul 2018, 6:07:35 UTC - in response to Message 35713.  

what I found out was that the tasks from the new version yield considerably less credit points compared to before. Regardless whether in 1-core or multicore mode.
Laurence, is this intentionally, or rather by mistake?
ID: 35724 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,466,562
RAC: 103,962
Message 35732 - Posted: 2 Jul 2018, 6:23:00 UTC - in response to Message 35713.  
Last modified: 2 Jul 2018, 6:23:49 UTC

what I found out was that the tasks from the new version yield considerably less credit points compared to before. Regardless whether in 1-core or multicore mode.
interestingly enough, two of my rigs (old ones, with really old CPUs) which run on Windows 32-bit systems, using
Therory Simulation v263.50 (vbox32)
yield markedly better credit results than the high-end machines running on 64-bit, using
Theory Simulation v263.60 (vbox64_mt_mcore)
ID: 35732 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,189,247
RAC: 104,501
Message 35734 - Posted: 2 Jul 2018, 7:25:10 UTC
Last modified: 2 Jul 2018, 7:25:35 UTC

This discussion about cobblestones can you read in older message-threads!
This thread is for new Theory-program!
Sorry, Erich.
ID: 35734 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,466,562
RAC: 103,962
Message 35735 - Posted: 2 Jul 2018, 8:02:57 UTC - in response to Message 35734.  

This thread is for new Theory-program!
And exactly that's the point: why does the new Theory-program yield so much less credit points?
I am just curious what's the reason behind it - not more and not less.
ID: 35735 · Report as offensive     Reply Quote
Profile Laurence
Project administrator
Project developer

Send message
Joined: 20 Jun 14
Posts: 372
Credit: 238,712
RAC: 0
Message 35736 - Posted: 2 Jul 2018, 8:36:33 UTC - in response to Message 35724.  

BOINC credit is a mystery even for us. The adapted credit algorithm just does not work in our view. This is an issue for may BOINC projects and is one of the reasons why they are forced to maintain their own forks of the code. For more details on this topic, you can read the issue in the github repository.
ID: 35736 · Report as offensive     Reply Quote
Jim1348

Send message
Joined: 15 Nov 14
Posts: 602
Credit: 24,371,321
RAC: 0
Message 35737 - Posted: 2 Jul 2018, 9:03:42 UTC - in response to Message 35736.  

The variability (or randomness) of BOINC credits is a continual subject for discussion at Rosetta.
http://boinc.bakerlab.org/rosetta/forum_thread.php?id=12133&postid=89199#89199
ID: 35737 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,466,562
RAC: 103,962
Message 35742 - Posted: 2 Jul 2018, 11:50:08 UTC - in response to Message 35736.  

BOINC credit is a mystery even for us. The adapted credit algorithm just does not work in our view. This is an issue for may BOINC projects and is one of the reasons why they are forced to maintain their own forks of the code. For more details on this topic, you can read the issue in the github repository.
thanks, Laurence, for your answer. The github repository articles/postings/comments are very interesting.
So the BOINC credit system indeed seems to be some kind of mystery (to say the least) :-)
ID: 35742 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Theory Application : New Version 263.60


©2024 CERN