3106

Message boards : Comments and discussion : 3106
Message board moderation

To post messages, you must log in.

AuthorMessage
overkill

Send message
Joined: 12 Jun 08
Posts: 56
Credit: 922,628
RAC: 0
Message 14572 - Posted: 24 Jun 2016, 12:59:36 UTC

Is this job too noisy to compare? I see WUs still pending with 3 results.
http://burp.renderfarming.net/workunit.php?wuid=2835622
ID: 14572 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
abHisHEK

Send message
Joined: 1 Mar 16
Posts: 10
Credit: 0
RAC: 0
Message 14573 - Posted: 24 Jun 2016, 17:49:00 UTC - in response to Message 14572.  

It is a Interior Scene with just a window with a sun lamp and a portal and a HDR and that is the only source of light,which explains why it is so noisy. The meshes are not too complex but about all of them use PBR shaders from poliigon
ID: 14573 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14574 - Posted: 25 Jun 2016, 9:25:04 UTC

Is the noise non-deterministic?
ID: 14574 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JZD

Send message
Joined: 30 Dec 11
Posts: 103
Credit: 3,498,559
RAC: 274
Message 14575 - Posted: 25 Jun 2016, 17:47:42 UTC

Why is increased "initial replication"? It are the units so bad that they have to be more of the same results?

ID: 14575 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
den777

Send message
Joined: 10 Dec 13
Posts: 2
Credit: 267,555
RAC: 0
Message 14576 - Posted: 25 Jun 2016, 19:01:31 UTC

Anybody have validated tasks from this session?
ID: 14576 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 7
Message 14579 - Posted: 26 Jun 2016, 6:55:25 UTC - in response to Message 14575.  

JZD wrote:
Why is increased "initial replication"? It are the units so bad that they have to be more of the same results?

Am wondering a bit myself, but it looks like the results differ from each other, otherwise they won't go into that pending state and need another result to compare.

den777 wrote:
Anybody have validated tasks from this session?

Doesn't look like it. There must be something seriously wrong with it when all tasks are going into "validation inconclusive".
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 14579 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
abHisHEK

Send message
Joined: 1 Mar 16
Posts: 10
Credit: 0
RAC: 0
Message 14580 - Posted: 26 Jun 2016, 8:27:12 UTC

is my blend file corrupt or wat ?? y is't it furnished rendering yet
ID: 14580 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14581 - Posted: 26 Jun 2016, 12:35:35 UTC - in response to Message 14580.  
Last modified: 26 Jun 2016, 13:10:58 UTC

Is the noise non-deterministic?
is my blend file corrupt or wat ?? y is't it furnished rendering yet

I take that as a yes. The session has been switched to "Best effort" rendering - we do not currently support non-deterministic results. In other words: The render must return the same result for each individual frame every time that frame is rendered. If you use shaders with custom random function you can use the random seed provided by Blender to make sure that the random functions give correct results every time.

This is likely to break the session.
ID: 14581 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 7
Message 14584 - Posted: 26 Jun 2016, 14:05:26 UTC - in response to Message 14581.  
Last modified: 26 Jun 2016, 14:16:43 UTC

The session has been switched to "Best effort" rendering - we do not currently support non-deterministic results. In other words: The render must return the same result for each individual frame every time that frame is rendered.

Great, looks like this made the majority of tasks invalid when I look over several users results.
Not sure if the results are really useless, but I highly doubt it.
I guess this calls for PayDay...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 14584 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14587 - Posted: 27 Jun 2016, 20:27:16 UTC

That is one of the unfortunate issues with the way the validation works in "Best effort"-mode, we have to pick an arbitrary result for BOINC to continue. Of course yes, the actual data is kept for Payday to use - and in cases like this it is pretty good at granting full credit to everyone.
ID: 14587 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Achim

Send message
Joined: 17 May 05
Posts: 183
Credit: 2,642,713
RAC: 0
Message 14589 - Posted: 30 Jun 2016, 6:47:40 UTC - in response to Message 14587.  

I've got the feeling the remaining ones, still need to validate.
I still see the validation inconclusive in the remaining WUs.
Could you please check?
ID: 14589 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14590 - Posted: 1 Jul 2016, 14:12:33 UTC - in response to Message 14589.  

You are absolutely right, trying to fix the remaining ones right now
ID: 14590 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
overkill

Send message
Joined: 12 Jun 08
Posts: 56
Credit: 922,628
RAC: 0
Message 14645 - Posted: 24 Jul 2016, 13:53:26 UTC

Has PayDay been run for this job yet? Is it something that has to be run manually, or only runs at a specific time (monthly/quarterly)?

Here's a task with 10 invalid results (for reference)
http://burp.renderfarming.net/workunit.php?wuid=2843383
ID: 14645 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 7
Message 14648 - Posted: 24 Jul 2016, 18:33:31 UTC - in response to Message 14645.  

Has PayDay been run for this job yet? Is it something that has to be run manually, or only runs at a specific time (monthly/quarterly)?

PayDay needs to be run manually from Janus, and it is not that often that he does it. Could be that it was only two times last year - and I can't rememeber that he already did it this year.
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 14648 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
JZD

Send message
Joined: 30 Dec 11
Posts: 103
Credit: 3,498,559
RAC: 274
Message 14655 - Posted: 27 Jul 2016, 12:42:30 UTC - in response to Message 14648.  

Could be that it was only two times last year - and I can't rememeber that he already did it this year.

I dare to oppose it. Last payday was more than two years ago, when the project ended Sunflowers. Thread Payday. http://burp.renderfarming.net/forum_thread.php?id=2101&postid=12378#12378
ID: 14655 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14658 - Posted: 28 Jul 2016, 15:52:49 UTC
Last modified: 28 Jul 2016, 15:54:15 UTC

The next Payday run has actually been running in the background for a few days now (just the analysis phase, not the credit granting phase). I didn't post about it yet because it will be mid-august or later before there is enough time to review the results.

It was delayed a bit because the code had to be updated to utilize the new version of the MySQL database plugin that we've switched to recently, updated to run alongside the project in normal non-maintenance mode and because the new GPU test sessions had to be filtered out of the statistical analysis.

So far it looks like the biggest Payday ever.
ID: 14658 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 7
Message 14663 - Posted: 28 Jul 2016, 18:05:01 UTC - in response to Message 14658.  

So far it looks like the biggest Payday ever.

That already sounds very promising.
I know there are quite a few sessions to be corrected but how much exactly are it at this point?
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 14663 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4570
Credit: 2,100,463
RAC: 8
Message 14664 - Posted: 29 Jul 2016, 10:52:02 UTC

No numbers yet, that will have to wait for mid Aug
ID: 14664 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Comments and discussion : 3106