Intermittent out of resources error clEnqueueReadBuffer failed

[boulderZ]

I can run multiple times and sometimes get error:
pyopencl.RuntimeError: clEnqueueReadBuffer failed: out of resources
after one or more iterations using the bike_a.png and bike_b.png files and main in demo mode straight from github. Sometimes it runs to completion and sometimes it stops part way through with above error.

Output on fail case:

Please wait a few seconds...
('file 0 = ', 'bike_a.png', ' file 1 = ', 'bike_b.png')
('img shape = ', (175, 239, 3))
('iteration', 1)
('mean block difference:', 11.847692226259401)
('iteration', 2)
('mean block difference:', 1.4053478590557273)
('iteration', 3)
('mean block difference:', 1.0751794090974331)
('iteration', 4)
('mean block difference:', 0.92336860183560998)
('iteration', 5)
('mean block difference:', 0.84552383540750564)
Traceback (most recent call last):
  File "main.py", line 195, in <module>
    patchmatch.match(files,Demo=True)
  File "main.py", line 186, in match
    self.execute();
  File "main.py", line 116, in execute
    cl.enqueue_read_buffer(self.queue, self.outputBuf, c).wait()
  File "/usr/lib/python2.7/dist-packages/pyopencl/__init__.py", line 860, in new_func
    return func(*args, **kwargs)
pyopencl.RuntimeError: clEnqueueReadBuffer failed: out of resources

When it runs successfully I get the following output:

Please wait a few seconds...
('file 0 = ', 'bike_a.png', ' file 1 = ', 'bike_b.png')
('img shape = ', (175, 239, 3))
('iteration', 1)
('mean block difference:', 11.847692226259401)
('iteration', 2)
('mean block difference:', 1.4053478590557273)
('iteration', 3)
('mean block difference:', 1.0751713650301336)
('iteration', 4)
('mean block difference:', 0.92228370782616687)
('iteration', 5)
('mean block difference:', 0.84449662336199371)
('iteration', 6)
('mean block difference:', 0.78995643332834575)
('iteration', 7)
('mean block difference:', 0.75182238942455692)
('iteration', 8)
('mean block difference:', 0.72269540056274861)
('iteration', 9)
('mean block difference:', 0.70064447195375668)
('iteration', 10)
('mean block difference:', 0.68212381211217721)
('iteration', 11)
('mean block difference:', 0.6671923072248922)
('iteration', 12)
('mean block difference:', 0.65498817675593513)
('iteration', 13)
('mean block difference:', 0.64466970594863771)
('iteration', 14)
('mean block difference:', 0.63470502613937019)
('iteration', 15)
('mean block difference:', 0.62660883279387825)
('iteration', 16)
('mean block difference:', 0.61789773527170389)
('iteration', 17)
('mean block difference:', 0.61108444301768117)
('iteration', 18)
('mean block difference:', 0.60394420541486304)
('iteration', 19)
('mean block difference:', 0.59827319833725168)
('iteration', 20)
('mean block difference:', 0.59298303466078317)
('randomfill', ':', 0.003, 'seconds')
('execute', ':', 0.137, 'seconds')
('init', ':', 0.519, 'seconds')
Done.

[abiusx]

Hmm.... I actually have no idea what's going on. I haven't touch this code for 3 years. Please let me know if you found out something else about it. Regards -A

…

On Aug 2, 2017, at 6:00 PM, Zak Keirn ***@***.***> wrote: I can run multiple times and sometimes get error: pyopencl.RuntimeError: clEnqueueReadBuffer failed: out of resources after one or more iterations using the bike_a.png and bike_b.png files and main in demo mode straight from github. Sometimes it runs to completion and sometimes it stops part way through with above error. Output on fail case: Please wait a few seconds... ('file 0 = ', 'bike_a.png', ' file 1 = ', 'bike_b.png') ('img shape = ', (175, 239, 3)) ('iteration', 1) ('mean block difference:', 11.847692226259401) ('iteration', 2) ('mean block difference:', 1.4053478590557273) ('iteration', 3) ('mean block difference:', 1.0751794090974331) ('iteration', 4) ('mean block difference:', 0.92336860183560998) ('iteration', 5) ('mean block difference:', 0.84552383540750564) Traceback (most recent call last): File "main.py", line 195, in <module> patchmatch.match(files,Demo=True) File "main.py", line 186, in match self.execute(); File "main.py", line 116, in execute cl.enqueue_read_buffer(self.queue, self.outputBuf, c).wait() File "/usr/lib/python2.7/dist-packages/pyopencl/__init__.py", line 860, in new_func return func(*args, **kwargs) pyopencl.RuntimeError: clEnqueueReadBuffer failed: out of resources When it runs successfully I get the following output: Please wait a few seconds... ('file 0 = ', 'bike_a.png', ' file 1 = ', 'bike_b.png') ('img shape = ', (175, 239, 3)) ('iteration', 1) ('mean block difference:', 11.847692226259401) ('iteration', 2) ('mean block difference:', 1.4053478590557273) ('iteration', 3) ('mean block difference:', 1.0751713650301336) ('iteration', 4) ('mean block difference:', 0.92228370782616687) ('iteration', 5) ('mean block difference:', 0.84449662336199371) ('iteration', 6) ('mean block difference:', 0.78995643332834575) ('iteration', 7) ('mean block difference:', 0.75182238942455692) ('iteration', 8) ('mean block difference:', 0.72269540056274861) ('iteration', 9) ('mean block difference:', 0.70064447195375668) ('iteration', 10) ('mean block difference:', 0.68212381211217721) ('iteration', 11) ('mean block difference:', 0.6671923072248922) ('iteration', 12) ('mean block difference:', 0.65498817675593513) ('iteration', 13) ('mean block difference:', 0.64466970594863771) ('iteration', 14) ('mean block difference:', 0.63470502613937019) ('iteration', 15) ('mean block difference:', 0.62660883279387825) ('iteration', 16) ('mean block difference:', 0.61789773527170389) ('iteration', 17) ('mean block difference:', 0.61108444301768117) ('iteration', 18) ('mean block difference:', 0.60394420541486304) ('iteration', 19) ('mean block difference:', 0.59827319833725168) ('iteration', 20) ('mean block difference:', 0.59298303466078317) ('randomfill', ':', 0.003, 'seconds') ('execute', ':', 0.137, 'seconds') ('init', ':', 0.519, 'seconds') Done. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABVjW4oyC_U-FT7WhkSIZsums6ll5Lj8ks5sUPF9gaJpZM4Ortqs>.

[boulderZ]

I will. Did you run it on any stereo image pairs for depth?

[abiusx]

I believe I did, but I don't remember any details. -A

…

On Aug 2, 2017, at 6:26 PM, Zak Keirn ***@***.***> wrote: I will. Did you run it on any stereo image pairs for depth? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABVjW6b2wrthgmcTq0AJDXffBZBllQSwks5sUPeugaJpZM4Ortqs>.

[boulderZ]

Does this implement the slanted windows version stereo patchmatch ?

[abiusx]

I'm not sure but I don't think so. It's for the original patchmatch algorithm. -A

…

On Aug 2, 2017, at 6:49 PM, Zak Keirn ***@***.***> wrote: Does this implement the slanted windows version stereo patchmatch <https://www.microsoft.com/en-us/research/wp-content/uploads/2011/01/PatchMatchStereo_BMVC2011_6MB.pdf> ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#4 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABVjW92jlOs8qsNNAmPIyS1W3iRXDqnhks5sUPzcgaJpZM4Ortqs>.

[boulderZ]

Thanks!