climateprediction.net home page
Batch 564 - perturbed physics

Batch 564 - perturbed physics

Message boards : Number crunching : Batch 564 - perturbed physics
Message board moderation

To post messages, you must log in.

AuthorMessage
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 56098 - Posted: 16 Apr 2017, 1:54:39 UTC

This is the main batch that was tested as batch 559.
More than the usual number of INVALID THETA should be expected, as these tasks are going to get slammed around a bit to see what happens.
ID: 56098 · Report as offensive     Reply Quote
Eirik Redd

Send message
Joined: 31 Aug 04
Posts: 391
Credit: 219,888,554
RAC: 1,481,373
Message 56100 - Posted: 16 Apr 2017, 6:57:22 UTC - in response to Message 56098.  

Thanks for the heads-up. Looks like these will run longer (if they finish) than the earlier recent 56* batches.
ID: 56100 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,377,675
RAC: 3,657
Message 56104 - Posted: 16 Apr 2017, 8:41:56 UTC

More than the usual number of INVALID THETA should be expected, as these tasks are going to get slammed around a bit to see what happens.


On this box under Linux, all three I got failed with sigsegv segmentation error, two had previously failed on other boxes, one with invalid theta the other with a segmentatin fault.
ID: 56104 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 56105 - Posted: 16 Apr 2017, 9:18:57 UTC - in response to Message 56104.  

Hi Dave

The part of your post where you mention "invalid theta" and "segmentation fault" close together has triggered a thought:

I wonder if SIGSEGV on Linux is the same as INVALID THETA on Windows?
Which triggered another thought:
Have I ever seen the two mentioned together in the same error listing?
(I don't think that I have.)

When the weekend is over, I'm going to mention this to Oxford.
ID: 56105 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,377,675
RAC: 3,657
Message 56107 - Posted: 16 Apr 2017, 9:38:38 UTC - in response to Message 56105.  

Hi Les,

possibly but I am sure I have had "invalid theta" on native linux tasks in the past. Not sure about segmentation violation on windows? The tasks were not failing at the same points either, one one my machine ran for longer than the previous one (could be because mine is quite slow) but on another task the windows machine ran for longer and it is quite a lot faster than this box.

However going through my tasks from the other place, invalid theta has only been under wine and sigsegv only on Linux so even if my historical memory is correct, your guess may be right with more recent builds for different platforms.
ID: 56107 · Report as offensive     Reply Quote
Desti

Send message
Joined: 6 Aug 04
Posts: 124
Credit: 9,195,838
RAC: 0
Message 56111 - Posted: 16 Apr 2017, 15:35:48 UTC

This failed an Windows with invalid theta and on Linux with segfault:

https://www.cpdn.org/cpdnboinc/workunit.php?wuid=11005566
https://www.cpdn.org/cpdnboinc/workunit.php?wuid=11006391
Linux Users Everywhere @ BOINC
ID: 56111 · Report as offensive     Reply Quote
Profile JIM

Send message
Joined: 31 Dec 07
Posts: 1152
Credit: 22,053,321
RAC: 4,417
Message 56114 - Posted: 16 Apr 2017, 19:13:20 UTC

Those starting parameter sets must be way out there. Out of 9 batch 564 tasks I have started only 1 has made it as far as 2 hours. If they are all like this it won’t take long to dispose of them, even if they are reissued 3 or 4 times.

The only real problem is this could be hard on Crunchers with low bandwidth caps. They might want to deselect the hadam3ps tasks until they are out of the system.
ID: 56114 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,403,322
RAC: 5,085
Message 56115 - Posted: 16 Apr 2017, 20:12:23 UTC

I must be incredibly lucky. I'm running 6, 4 in linux and 2 in Wine/Windows. So far all are progressing along nicely, the Wine/Windows tasks are past 60% and the linux ones are past 20%.
ID: 56115 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 56116 - Posted: 16 Apr 2017, 21:06:08 UTC

2 on Linux at 18 hours, plus 1 last one just started, 2 on Wine at 24 hours, and 2 on Win 7 at just under 4 hours.

Looks like luck of the drawer.
I think that the way it works, is that the generator creates a task from several lists of values, drops it in the bin, increments one of the lists to get the next set of values, drops that in the bin, etc.
When the required number of tasks has been created, it stops.

So if a computer asks for a number of tasks at once during a "quiet period", it may get all of the tasks with only slightly different values.
ID: 56116 · Report as offensive     Reply Quote
Profile JIM

Send message
Joined: 31 Dec 07
Posts: 1152
Credit: 22,053,321
RAC: 4,417
Message 56118 - Posted: 16 Apr 2017, 22:52:45 UTC

I figured that there probably would be stable regions in the data sets that yield viable climates scattered like islands through a vast sea of variables that don’t work together. I guess that I just haven’t downloaded from one of those islands.
ID: 56118 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4314
Credit: 16,377,675
RAC: 3,657
Message 56258 - Posted: 19 May 2017, 10:49:39 UTC - in response to Message 56118.  

JUst had two of these as re-treads that have failed with

<stderr_txt>
forrtl: severe (17): syntax error in NAMELIST input, unit 5, file /var/lib/boinc-client/projects/climateprediction.net/hadcm3s_83ys_201412_120_564_011007447/jobs/climate.cpdc, line 396, position 20
Image PC Routine Line Source

after about a few seconds.
ID: 56258 · Report as offensive     Reply Quote

Message boards : Number crunching : Batch 564 - perturbed physics

©2024 climateprediction.net