Again error during computing - what can I do?
16 Sep 2020 14:19:39 UTC
Topic 223504
(moderation:
I hat this type of error some weeks before. Now I tried again and got the following error message:
16.09.2020 12:15:41 | Einstein@Home | Computation for task h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0 finished
16.09.2020 12:15:41 | Einstein@Home | Output file h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0_0 for task h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0 absent
16.09.2020 12:15:41 | Einstein@Home | Output file h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0_1 for task h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0 absent
16.09.2020 12:15:41 | Einstein@Home | Output file h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0_2 for task h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0 absent
16.09.2020 12:15:43 | Einstein@Home | Starting task LATeah1065L47_484.0_0_0.0_30207751_0
16.09.2020 12:15:44 | Einstein@Home | Started upload of h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0_3
16.09.2020 12:15:46 | Einstein@Home | Finished upload of h1_0345.25_O2C02Cl4In0__O2MDFS2_Spotlight_345.70Hz_1323_0_3
16.09.2020 12:16:46 | Einstein@Home | Project requested delay of 60 seconds
Another work unit has completed but is still waiting for validation:
16.09.2020 12:53:58 | Einstein@Home | Computation for task LATeah1065L47_484.0_0_0.0_30207751_0 finished
16.09.2020 12:54:01 | Einstein@Home | Started upload of LATeah1065L47_484.0_0_0.0_30207751_0_0
16.09.2020 12:54:01 | Einstein@Home | Started upload of LATeah1065L47_484.0_0_0.0_30207751_0_1
16.09.2020 12:54:03 | Einstein@Home | Finished upload of LATeah1065L47_484.0_0_0.0_30207751_0_0
16.09.2020 12:54:03 | Einstein@Home | Finished upload of LATeah1065L47_484.0_0_0.0_30207751_0_1
16.09.2020 12:54:07 | Einstein@Home | Project requested delay of 60 seconds
What can I do to prevent the error in h1_... tasks ?
thanks for help
Jochen
Hi ! Stderr output of that
)
Hi !
Stderr output of that failed task includes this text: "Processing FFT failed: CL_MEM_OBJECT_ALLOCATION_FAILURE"
That is believed to be caused by a task requiring more memory than what the GPU can provide. Your card has 2 GB and not all of that is available for Einstein to start with. Some of the Gravitational Wave tasks have required more than 2 GB. That particular task that failed has so called DF value of 0.45. That is quite high. Some tasks with DF of 0.40 have used plenty of memory already. It could be that this task wanted even a larger chunk.
Is your computer doing anything else while crunching these tasks ?
Maybe you could try yet another GW task and see if the server sent a task from different data area, possibly with smaller requirements this time.
Hi Richie, Thank for this
)
Hi Richie,
Thank for this quick and useful answer. indeed, I have a CPU with 4 processors and BOINC is limited to 3 of them + NVIDIA. I run a lot of other processes but there are no other tasks using the GPU.
What do you mean by "GW task" `? Graphic work ?
I'll try.
Dr_Mabuse wrote:indeed, I
)
Okay, I can't think of any reason why that shouldn't work.
Gravitation Wave... (Gravitational Wave search O2 Multi-Directional GPU)... those "h1_... " tasks.
Oh, one thing!
I noticed now that your GW tasks have been v2.07. There is Beta version v2.09 of this GW GPU application available that solves some problems (some of those memory issues too, if I remember right).
Please visit https://einsteinathome.org/account/prefs/project and set "Run test applications" under Beta settings to YES. Save and force an update for Einstein.
Then your computer should receive v2.09 tasks (instead of v2.07) and let's see how they will run.
Yeah it looks like 309 Hz
)
Yeah it looks like 309 Hz with 0.35 DF works well for your card now even if they were v2.07 tasks. The few recent tasks that failed were probably looking for a large amount of GPU memory but those current tasks fit in.