someone explain these wu\'s please

Message boards : Number crunching : someone explain these wu\'s please

To post messages, you must log in.

AuthorMessage
Profile nubz

Send message
Joined: 18 Sep 05
Posts: 3
Credit: 142,568
RAC: 519
Message 22770 - Posted: 17 Aug 2006, 19:00:37 UTC

http://boinc.bakerlab.org/rosetta/result.php?resultid=32329251
This process generated 59 decoys from 59 attempts
Granted credit 21.7816396684088
Granted work credit 1.2804942607273

and this one:

http://boinc.bakerlab.org/rosetta/result.php?resultid=32329252
This process generated 42 decoys from 42 attempts
Granted credit 21.5844895146452
Granted work credit 26.9278520568427

big difference.
ID: 22770 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [B^S] thierry@home
Avatar

Send message
Joined: 17 Sep 05
Posts: 182
Credit: 281,902
RAC: 0
Message 22771 - Posted: 17 Aug 2006, 19:09:36 UTC
Last modified: 17 Aug 2006, 19:10:11 UTC

The first one is the result of a error in some scripts this morning (or evening) producing a low result in the Granted Work Credit (only for a few WUs). It\'s OK in the second WU.

The Granted credit is the credit you receive for a given WU. The Granted work credit is the \'new\' credit system which is being tested.
ID: 22771 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Tallbill

Send message
Joined: 23 Jul 06
Posts: 12
Credit: 101,854
RAC: 0
Message 22774 - Posted: 17 Aug 2006, 19:19:48 UTC

So at this point, the Granted work credit means nothing, but eventually all of our points will turn into that? Or they still havn\'t decided?
ID: 22774 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [B^S] thierry@home
Avatar

Send message
Joined: 17 Sep 05
Posts: 182
Credit: 281,902
RAC: 0
Message 22776 - Posted: 17 Aug 2006, 19:22:56 UTC

Yes it\'s only a test.
There\'s no decision until the project team find an agreement with users.
ID: 22776 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Scribe
Avatar

Send message
Joined: 2 Nov 05
Posts: 284
Credit: 157,359
RAC: 0
Message 22779 - Posted: 17 Aug 2006, 19:27:49 UTC

If it was a test then it should have been kept and tested in the labs, not out in the wild. all it has succeeded in doing is driving out a lot of us! You say with agreement of the users, who, the volatile ones or the ordinary ones like the vast majority....it should havebeen left exactly as it was!
ID: 22779 · Rating: -1 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile nubz

Send message
Joined: 18 Sep 05
Posts: 3
Credit: 142,568
RAC: 519
Message 22780 - Posted: 17 Aug 2006, 19:36:32 UTC - in response to Message 22771.  

The first one is the result of a error in some scripts this morning (or evening) producing a low result in the Granted Work Credit (only for a few WUs). It\'s OK in the second WU.

The Granted credit is the credit you receive for a given WU. The Granted work credit is the \'new\' credit system which is being tested.



ok - thanks for the info.
ID: 22780 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dgnuff
Avatar

Send message
Joined: 1 Nov 05
Posts: 347
Credit: 24,023,248
RAC: 4
Message 22859 - Posted: 18 Aug 2006, 0:38:38 UTC - in response to Message 22779.  

If it was a test then it should have been kept and tested in the labs, not out in the wild. all it has succeeded in doing is driving out a lot of us! You say with agreement of the users, who, the volatile ones or the ordinary ones like the vast majority....it should havebeen left exactly as it was!


Sometimes the only way to really test something is to throw it to the wolves. As a case in poin, I\'m working on some code that is a software simulation of a broken floating point processor that\'s sufficiently different from IEEE that the FPU in the host system (a PPC) can\'t be used.

So I had to write a FP add using pure integer ops, tear the two inputs apart into sign, exponent and mantissa, try to duplicate the hardware (for which we have NO documentation at all), and produce an answer.

After a week or so, I knew enough to get something that seemed to work. So we set it up for a soak test by throwing random numbers at it.

It took a week and three days running 24 / 7, about 5 million tests a second, before one pathological case caused a breakage.

I\'m not saying that the new credit system is quite this squirrely, but at the same time, it does need a hard soak test before going live.

That\'s what it\'s getting. I\'d sugest that rather than removing it, there could have been a better explanation of exactly whaddinhell was going on.

The problem is people tend to see low numbers and immediately think the whole credit system is about to crash, burn and go down in flames. Well, despite what Chicken Little may have believed, the sky is not falling.
ID: 22859 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Ulrich Metzner
Avatar

Send message
Joined: 17 Sep 05
Posts: 22
Credit: 255,680
RAC: 0
Message 22866 - Posted: 18 Aug 2006, 1:33:17 UTC
Last modified: 18 Aug 2006, 1:33:35 UTC

Maybe this one has to be added to the list:
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=28344343
There is a discrepancy in claimed and granted credit.
greetz, Uli

ID: 22866 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1001
Credit: 2,862,848
RAC: 3,364
Message 22873 - Posted: 18 Aug 2006, 1:51:56 UTC

I fixed the granted work credit values in the results. Please post any significant discrepancies here.
ID: 22873 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : someone explain these wu\'s please



©2017 University of Washington
http://www.bakerlab.org