OpenCL AstroPulse crash after processing completion - write here.

If you see computation errors with OpenCL AstroPulse (AP) application and in stderr of task you see that computations were finished (that is, number of found pulses printed in stderr, counters are printed in stderr and after that debug dump occured), please, report in this thread with relevant pecularities of your setup.

Example of such crash:

(in bold - parts that confirm that scientific processing of task fully completed before crash)

single pulses: 0
repetitive pulses: 30
percent blanked: 9.53
class T_remove_radar: total=3.91e+009, N=1, <>=3.91e+009, min=3.91e+009, max=3.91e+009
class T_main_loop_L1: total=5.04e+013, N=111, <>=4.54e+011, min=4.51e+011, max=4.58e+011
class T_FFT_forward: total=2.64e+010, N=909312, <>=2.91e+004, min=7.58e+003, max=3.25e+007
class T_remove_radar_randomize: total=1.94e+012, N=1817736, <>=1.07e+006, min=2.88e+002, max=2.75e+007
class T_build_chirp_table: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_DataWrite: total=1.82e+009, N=88800, <>=2.05e+004, min=3.27e+003, max=4.65e+005
class T_DataWrite_ns: total=0, N=0, <>=0, min=0 max=0
class T_oclReadBuf: total=3.14e+007, N=909312, <>=3.40e+001, min=2.40e+001, max=3.38e+005
class T_ChirpWrite: total=0.00e+000, N=0, <>=0.00e+000, min=1.84e+019, max=0.00e+000
class T_ChirpWrite_ns: total=0, N=0, <>=0, min=0 max=0
class T_dechirp: total=2.45e+010, N=909312, <>=2.70e+004, min=1.01e+004, max=6.90e+006
class Dechirp_ns: total=0, N=0, <>=0, min=0 max=0
class Half_ns: total=0, N=0, <>=0, min=0 max=0
class T_PC_single_pulse_kernel_FFA_update: total=4.83e+013, N=909312, <>=5.31e+007, min=5.16e+007, max=1.28e+008
class PC_ns: total=0, N=0, <>=0, min=0 max=0
class T_oclReadBuf: total=3.14e+007, N=909312, <>=3.40e+001, min=2.40e+001, max=3.38e+005
class T_oclWriteBuf: total=1.86e+009, N=88800, <>=2.09e+004, min=3.38e+003, max=4.66e+005
class T_FFT_inverse: total=1.12e+010, N=909312, <>=1.23e+004, min=5.61e+003, max=6.87e+006
class T_ffa: total=2.08e+009, N=1, <>=2.08e+009, min=2.08e+009, max=2.08e+009
class T_GPU_buffer_read_backs: total=0, N=0, <>=0, min=0 max=0
OCL_ZERO_COPY USE_OPENCL OPENCL_WRITE USE_INCREASED_PRECISION SMALL_CHIRP_TABLE COMBINED_DECHIRP_KERNEL
rev 1316


Unhandled Exception Detected...

- Unhandled Exception Record -
Reason: Access Violation (0xc0000005) at address 0x0040A1FA read attempt to address 0x00399E0C

Engaging BOINC Windows Runtime Debugger...
____________
News about SETI opt app releases: https://twitter.com/Raistmer

IF you experience this issue please upgrade to these apps:
ATi: https://dl.dropbox.com/u/60381958/AP6_win_x86_SSE2_OpenCL_ATI_r1764.7z
NV: https://dl.dropbox.com/u/60381958/AP6_win_x86_SSE2_OpenCL_NV_r1764.7z

Watch for next message in stderr:

ERROR: some exception inside XXXXXXX, doing hard termination...

I don't think I've had any crashes with AP before - or if I have, then only very occasionally. What's in the new revisions? Extra debugging info?
____________
Soli Deo Gloria

I don't think I've had any crashes with AP before - or if I have, then only very occasionally. What's in the new revisions? Extra debugging info?

I definitely had two of these, but I reset them and recrunched them with no problems on second run. Validated too.
Was weird error, I thought it was something wrong with my overclocked hardware.

I'm struggling more with the sudden stop of processing(HD 5770). Happens once a day or two. Mostly, it is enough to restart BOINC to get it run again. I even developed special application, which detects which particular GPU is idle, but has AstroPulse process, assigned to it and this application just restarts BOINC. The problem is, these restarts bring me BSOD from time to time, caused by ATI driver. Tried many different APP SDKs, but none solved the problem completely. There even was one task, which stopped being processed no matter now many times I restart BOINC. Just couldn't continue, stops at the same percentage. Rescheduled it to CPU to get it done.

I havent saved anything unfortunately, bcs, as I said, I thought it was solely on my side.
____________

IF you experience this issue please upgrade to these apps:
ATi: https://dl.dropbox.com/u/60381958/AP6_win_x86_SSE2_OpenCL_ATI_r1764.7z
NV: https://dl.dropbox.com/u/60381958/AP6_win_x86_SSE2_OpenCL_NV_r1764.7z

Watch for next message in stderr:

ERROR: some exception inside XXXXXXX, doing hard termination...

Now that these tasks are being rerun and listed as a Success, they will be much more difficult to find. Much more time consuming than just looking for an Error Listing. Do you know if this rerun 'feature' is just in 7.0.45 or are all the newer versions going to be this way? No more wasted time, the days of a wasted task appear to be over. I just installed the new App, should be finished in about 20 minutes. It would be nice if Uploads were working...
:-)

I definitely had two of these, but I reset them and recrunched them with no problems on second run. Validated too.
Was weird error, I thought it was something wrong with my overclocked hardware.

I'm struggling more with the sudden stop of processing(HD 5770). Happens once a day or two. Mostly, it is enough to restart BOINC to get it run again. I even developed special application, which detects which particular GPU is idle, but has AstroPulse process, assigned to it and this application just restarts BOINC. The problem is, these restarts bring me BSOD from time to time, caused by ATI driver. Tried many different APP SDKs, but none solved the problem completely. There even was one task, which stopped being processed no matter now many times I restart BOINC. Just couldn't continue, stops at the same percentage. Rescheduled it to CPU to get it done.

I havent saved anything unfortunately, bcs, as I said, I thought it was solely on my side.

Thanks for info. And very good feature indeed. To lose complete task when it actually was finished already was not very good.
____________
News about SETI opt app releases: https://twitter.com/Raistmer

Try to free more cores. If enough CPU is free I see no stuck, if not enough - I got such stuck tasks too.

... not here.

Hm.... no exception interception occured.
Either C++ exception handling doesn't work there or exception happened besides of watched area.
____________
News about SETI opt app releases: https://twitter.com/Raistmer

09:21:14 (2264): called boinc_finish