Astropulse work units locking up


log in

Advanced search

Message boards : Number crunching : Astropulse work units locking up

Author Message
Kevster
Send message
Joined: 11 Jan 01
Posts: 33
Credit: 1,311,513
RAC: 372
Canada
Message 1427707 - Posted: 12 Oct 2013, 19:06:00 UTC

I've been out of the game for quite a while, so this may be a known issue, or not an issue. Every time I try to crunch an astropulse work unit, it gets to approx. 22% complete, then seems to lock up. The "elapsed time" continues onward (highest I've seen is about 80 hours), but the "progress" and "remaining time" doesn't change. This has happened on more than three different ap work units. All my non-ap work units munch along nicely. Any ideas?
____________

Kevster
Send message
Joined: 11 Jan 01
Posts: 33
Credit: 1,311,513
RAC: 372
Canada
Message 1427713 - Posted: 12 Oct 2013, 19:11:55 UTC

Oops, forgot to mention that it states (.133 CPU's + 1 ATI GPU)beside the work unit. I'm now thinking this isn't an astropulse issue, but a lack of my understanding of what (.133 CPU's + 1 ATI GPU) means. Still, why is this apparently locking up? I've never seen one go past approx. 22%.
____________

Profile MikeProject donor
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 25190
Credit: 34,787,961
RAC: 20,890
Germany
Message 1427794 - Posted: 12 Oct 2013, 20:45:08 UTC

Have you recently updated to win 8 ?
It seems you have some driver left overs.

____________

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2894
Credit: 6,625,497
RAC: 8,034
Bulgaria
Message 1427812 - Posted: 12 Oct 2013, 21:06:48 UTC - in response to Message 1427707.
Last modified: 12 Oct 2013, 21:54:45 UTC


Similar problem (task hang at ~21%) was discussed (17 Jun 2013) here:
http://setiathome.berkeley.edu/forum_thread.php?id=72035

Same parameters:
Windows 8: Core x64 Edition
AMD Radeon HD 7400/7500/7600 series (Trinity) CAL version 1.4.1741, 512MB
BOINC version 7.0.64 for windows_x86_64

The different - it was happening with SETI7 OpenCL r1843, not with Astropulse.

At the end of the long discussion the user solved the problem by installing older BOINC 7.0.28
http://setiathome.berkeley.edu/forum_thread.php?id=72035&postid=1385824#1385824


But the only AstroPulse task for computer 7067267 is one CPU task 'AstroPulse v6 v6.01' which is 'Aborted by user':
http://setiathome.berkeley.edu/results.php?hostid=7067267&offset=0&show_names=0&state=0&appid=12

Which contradicts with the "forgot to mention that it states (.133 CPU's + 1 ATI GPU) beside the work unit"


____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2894
Credit: 6,625,497
RAC: 8,034
Bulgaria
Message 1427844 - Posted: 12 Oct 2013, 22:30:03 UTC - in response to Message 1427812.

But the only AstroPulse task for computer 7067267 is one CPU task 'AstroPulse v6 v6.01' which is 'Aborted by user':
http://setiathome.berkeley.edu/results.php?hostid=7067267&offset=0&show_names=0&state=0&appid=12

Which contradicts with the "forgot to mention that it states (.133 CPU's + 1 ATI GPU) beside the work unit"

Which makes me believe the hang task for ATI GPU was not AstroPulse but 'SETI@home v7 v7.03 (opencl_ati5_sah)' like this:
http://setiathome.berkeley.edu/result.php?resultid=3124037807

It Restarted 3 times at ~22 percent (and after 'Run time 85,376.02' was 'Aborted by user'):

Running on device number: 0
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
OpenCL platform detected: Advanced Micro Devices, Inc.
BOINC assigns device 0
Info: BOINC provided device ID used

Build features: SETI7 Non-graphics OpenCL OCL_CHIRP3 FFTW AMD specific USE_SSE x86
CPUID: AMD A8-5500 APU with Radeon(tm) HD Graphics

Cache: L1=64K L2=2048K

CPU features: FPU TSC PAE CMPXCHG8B APIC SYSENTER MTRR CMOV/CCMP MMX FXSAVE/FXRSTOR SSE SSE2 HT SSE3
OpenCL-kernels filename : MultiBeam_Kernels_r1831.cl
ar=0.444557 NumCfft=191999 NumGauss= 1063223448 NumPulse= 114410226175 NumTriplet= 15762032885760

Currently allocated 145 MB for GPU buffers
Restarted at 22.04 percent.
Used GPU device parameters are:
Number of compute units: 4
Single buffer allocation size: 64MB
max WG size: 256
period_iterations_num=20


____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile Bill Walker
Avatar
Send message
Joined: 4 Sep 99
Posts: 3458
Credit: 2,213,573
RAC: 956
Canada
Message 1429011 - Posted: 15 Oct 2013, 23:05:02 UTC

I have a similar problem, but not on a GPU. Machine ID is 4630841, WU is
1332491653. This is stuck at 77% complete, but elapsed time continues to count up. Checking properties, this WU has been at about 12 hours CPU time for 2 days, although elapsed times is now over 47 hours. This machine gets shut down twice a day, to go to and from work. Anybody have any ideas?
____________

StickProject donor
Volunteer tester
Send message
Joined: 26 Feb 00
Posts: 94
Credit: 1,744,488
RAC: 660
United States
Message 1429064 - Posted: 16 Oct 2013, 2:20:55 UTC

A while ago, I reported several stuck ATI tasks (both V7 and AP) on this thread: Stuck AstroPulse v6 v6.06 (opencl_ati_100) task. I am now having similar problems with AP CPU tasks (AstroPulse v6 v6.01) on the same host. And, strangely enough, the problem with stuck GPU ATI tasks has pretty much gone away. As with the stuck ATI tasks, when an AstroPulse v6 v6.01 gets stuck, I am usually able to get it going again by restarting BOINC - but sometimes I also need to suspend all GPU tasks first and then restart BOINC.
____________

Profile ivan
Volunteer tester
Avatar
Send message
Joined: 5 Mar 01
Posts: 645
Credit: 147,694,895
RAC: 47,494
United Kingdom
Message 1429407 - Posted: 16 Oct 2013, 21:27:27 UTC - in response to Message 1429064.

A while ago, I reported several stuck ATI tasks (both V7 and AP) on this thread: Stuck AstroPulse v6 v6.06 (opencl_ati_100) task. I am now having similar problems with AP CPU tasks (AstroPulse v6 v6.01) on the same host. And, strangely enough, the problem with stuck GPU ATI tasks has pretty much gone away. As with the stuck ATI tasks, when an AstroPulse v6 v6.01 gets stuck, I am usually able to get it going again by restarting BOINC - but sometimes I also need to suspend all GPU tasks first and then restart BOINC.

If I had the time at the moment, I'd run this over the BOINC and s@h codebase.
http://phys.org/news/2013-10-dude-code.html
http://pdos.csail.mit.edu/~xi/papers/stack-sosp13.pdf
____________

merle van osdol
Send message
Joined: 23 Oct 02
Posts: 771
Credit: 1,762,568
RAC: 1,856
United States
Message 1435290 - Posted: 29 Oct 2013, 22:17:06 UTC - in response to Message 1429407.

I get the same hanging action for astropulse and v7 using a hd6570 gpu. I am using win 8.1. I hate to give up on seti AGAIN but I am not going to have any choice unless someone has a solution. I am using the latest BOINC. Any help?
Thanks. I will have to go back to collatz or moo!wrapper if not. Thanks for reading this.
____________
nO cOmon cEnTs - vote yes for freedom of speech

Profile BilBg
Volunteer tester
Avatar
Send message
Joined: 27 May 07
Posts: 2894
Credit: 6,625,497
RAC: 8,034
Bulgaria
Message 1435591 - Posted: 30 Oct 2013, 13:51:18 UTC - in response to Message 1435290.

I am using the latest BOINC. Any help?

http://setiathome.berkeley.edu/forum_thread.php?id=73008&postid=1427812#1427812


____________



- ALF - "Find out what you don't do well ..... then don't do it!" :)

Profile Carmine Ventura
Send message
Joined: 8 Sep 99
Posts: 32
Credit: 534,615
RAC: 82
United States
Message 1443154 - Posted: 16 Nov 2013, 20:34:59 UTC

Today I got my FIRST Astropulse work unit ! Its about time .
____________

Message boards : Number crunching : Astropulse work units locking up

Copyright © 2014 University of California