Gamma-ray pulsar binary search is seemingly in an infinite loop

Balazs Somogyi
Balazs Somogyi
Joined: 20 Mar 16
Posts: 5
Credit: 87418481
RAC: 0
Topic 198527

I have two Gamma-ray pulsar binary search #1 1.00 running, but they are both stuck around these numbers:
LATeah0001L_432.0_0_0.0_1089550_4: ~69.3% progress, ~08:27:00 elapsed, ~14:46:00 remaining
LATeah0001L_976.0_0_0.0_8986950_1: ~69.8% progress, ~08:34:00 elapsed, ~14:43:00 remaining

it still seems to calculate something, the values are increasing, but after a while jump back a little, so the values are arond the numbers given above.

I'm running it on OS X 10.10.5, BOINC 7.6.12.
Do you have any idea what's going on? Do I need to supply any additional info?

Cheers,
Balazs

jd
jd
Joined: 13 Mar 05
Posts: 36
Credit: 557477738
RAC: 85081

Gamma-ray pulsar binary search is seemingly in an infinite loop

I have the same problem many times. I also have the same problem on a GTX 750. When I see it, I just abort them, but it sure is a waste!

Balazs Somogyi
Balazs Somogyi
Joined: 20 Mar 16
Posts: 5
Credit: 87418481
RAC: 0

Is anyone reading this from

Is anyone reading this from the development team? Have you seen this behaviour? Do you have any plans to fix it?

It is a really great project, and I'm glad to be part of it, but I certainly don't want to waste computer cycles on useless computation.

Logforme
Logforme
Joined: 13 Aug 10
Posts: 332
Credit: 1714373961
RAC: 0

Your computer is hidden so I

Your computer is hidden so I can't see what kind of GPU you are using.
If it is an AMD GPU I (and some others) have on rare occasions seen that the task will "go on forever". Suspending the task and then starting it again will get the task working ok.

Balazs Somogyi
Balazs Somogyi
Joined: 20 Mar 16
Posts: 5
Credit: 87418481
RAC: 0

I made my computers visible,

I made my computers visible, both have AMD GPU.
I suspended and then restarted the tasks, but it didn't help. It won't work anyway, because I'm running it remotely on my other machine, and I can't check the state regularly.
Do you think disabling GPU in the Einstein@Home preferences will fix the issue?

Logforme
Logforme
Joined: 13 Aug 10
Posts: 332
Credit: 1714373961
RAC: 0

I'm sorry but I didn't read

I'm sorry but I didn't read your question correctly. You're not running any GPU tasks at all (since your OSX version is too new and your windows machine seems to not have an OpenCL capable driver?) so my suggestion won't work.
Sorry about the confusion

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5872
Credit: 117736618878
RAC: 34890305

RE: I made my computers

Quote:
I made my computers visible, both have AMD GPU.


Thanks for doing that. It makes it a lot easier to comment when we can see the basic hardware details and the types of tasks being crunched.

Quote:
I suspended and then restarted the tasks, but it didn't help. It won't work anyway, because I'm running it remotely on my other machine, and I can't check the state regularly.


I don't understand what you mean. What is the "it" in "It won't work anyway ..."? When you say, "running it remotely" do you mean that you are running a BOINC Manager one one machine and using it to remotely connect to a BOINC client on the second machine? If so, there's absolutely no reason why you can't remotely suspend and resume crunching using the manager on the local machine to control the client on the remote machine. If this is not what you are talking about, please explain.

In your original message, you mentioned a machine running OS X. That machine currently has 4 FGRPB1 tasks, two completed and returned and two 'in progress'. The two 'in progress' are the two you listed. Since they still haven't completed, you should try to completely reboot the machine. It's not really likely to be an issue with the app or the data as there are no widespread reports of this type of behaviour from other users.

Quote:
Do you think disabling GPU in the Einstein@Home preferences will fix the issue?


It's not a GPU issue (FGRPB1 tasks are CPU only) so disabling the GPU in preferences won't have any effect. Your machine appears to be a laptop and BOINC sees 8 virtual cores. If you are trying to run 8 CPU tasks on a laptop, it may well be overheating and the behaviour you are seeing might be due to that. You could try allowing BOINC to use say 50% of the cores to see if that would allow things to work better.

Cheers,
Gary.

AgentB
AgentB
Joined: 17 Mar 12
Posts: 915
Credit: 513211304
RAC: 0

RE: I made my computers

Quote:
I made my computers visible, both have AMD GPU.

As Gary says, thank you this helps.

You are running BOINC version 6.10.58 on Windows 7 - i'd suggest running a much later version if possible.

Balazs Somogyi
Balazs Somogyi
Joined: 20 Mar 16
Posts: 5
Credit: 87418481
RAC: 0

@Logforme: No problem, thanks

@Logforme: No problem, thanks for your answer!

Balazs Somogyi
Balazs Somogyi
Joined: 20 Mar 16
Posts: 5
Credit: 87418481
RAC: 0

RE: I don't understand what

Quote:
I don't understand what you mean. What is the "it" in "It won't work anyway ..."? When you say, "running it remotely" do you mean that you are running a BOINC Manager one one machine and using it to remotely connect to a BOINC client on the second machine? If so, there's absolutely no reason why you can't remotely suspend and resume crunching using the manager on the local machine to control the client on the remote machine. If this is not what you are talking about, please explain.


My windows machine is a couple of thousand miles away from me, so although it is technically possible to remote login to that, in fact I do it occasionally, I certainly can't keep an eye on the BOINC client on a daily basis.

Quote:
In your original message, you mentioned a machine running OS X. That machine currently has 4 FGRPB1 tasks, two completed and returned and two 'in progress'. The two 'in progress' are the two you listed. Since they still haven't completed, you should try to completely reboot the machine. It's not really likely to be an issue with the app or the data as there are no widespread reports of this type of behaviour from other users.


On the Mac there is no Einstein@Home project running at the moment for some reason (and none is even downloaded), only the other project I have is running (which is WorldCommunityGrid by the way), though the resource share is set to 50-50 percent. The two problematic tasks, which I mentioned in the original post, are currently suspended, but I will abort them if the problem will not be resolved. I will try rebooting first, of course.

Quote:
It's not a GPU issue (FGRPB1 tasks are CPU only) so disabling the GPU in preferences won't have any effect. Your machine appears to be a laptop and BOINC sees 8 virtual cores. If you are trying to run 8 CPU tasks on a laptop, it may well be overheating and the behaviour you are seeing might be due to that. You could try allowing BOINC to use say 50% of the cores to see if that would allow things to work better.


It is indeed a laptop with 8 cores. It is not overheating at all, I set the CPU usage to be at most 50% to avoid unwanted slowness, which I experienced when running with 100%. I never had any problem with WorldCommunityGrid, which I have been running on this laptop for more that half a year now.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.