Is your device set for exclusive mode? (test_multi_context is the relevant failures, all the others fail because they try to use a now-dead context.)
I suppose we should check for that on 2.2, and skip the multi_context test if the device is in exclusive mode, regardless of whether that's what the issue is here. Andreas On Sonntag 26 Juli 2009, Nicolas Pinto wrote: > On Sun, Jul 26, 2009 at 6:05 PM, Nicolas Pinto <[email protected]> wrote: > > Quick update using the latest git. I'm getting the same "invalid context" > > error. > > > > See error.log attached (obtained using py.test -v test_driver.py 2>&1 > > > error.log). > > > > Regards, > > > > On Sun, Jul 26, 2009 at 5:57 PM, Nicolas Pinto <[email protected]> wrote: > >> Andreas, > >> > >> I'm getting this error using pycuda-0.92, here is the output of py.test > >> -v: > >> > >> ======================================================================== > >>====== test session starts > >> ======================================================================== > >>====== python: platform linux2 -- Python 2.5.1 -- /usr/bin/python -- > >> pytest-1.0.0b8 > >> test object 1: /home/ac/gpu052/src/pycuda-0.92/test > >> > >> test_abstract_array.py:47: TestAbstractArray.test_abs PASS > >> test_abstract_array.py:140: TestAbstractArray.test_addition_array FAIL > >> test_abstract_array.py:158: TestAbstractArray.test_addition_scalar FAIL > >> test_abstract_array.py:63: TestAbstractArray.test_arange PASS > >> test_abstract_array.py:257: TestAbstractArray.test_divide_array FAIL > >> test_abstract_array.py:234: TestAbstractArray.test_divide_scalar FAIL > >> test_abstract_array.py:86: TestAbstractArray.test_len FAIL > >> test_abstract_array.py:92: TestAbstractArray.test_multiply FAIL > >> test_abstract_array.py:122: TestAbstractArray.test_multiply_array FAIL > >> test_abstract_array.py:19: TestAbstractArray.test_pow_array FAIL > >> test_abstract_array.py:35: TestAbstractArray.test_pow_number FAIL > >> test_abstract_array.py:72: TestAbstractArray.test_reverse FAIL > >> test_abstract_array.py:182: TestAbstractArray.test_substract_array FAIL > >> test_abstract_array.py:208: TestAbstractArray.test_substract_scalar FAIL > >> test_driver.py:133: TestCuda.test_2d_texture PASS > >> test_driver.py:300: TestCuda.test_3d_texture PASS > >> test_driver.py:251: TestCuda.test_bitlog PASS > >> test_driver.py:109: TestCuda.test_gpuarray PASS > >> test_driver.py:231: TestCuda.test_large_smem PASS > >> test_driver.py:20: TestCuda.test_memory PASS > >> test_driver.py:271: TestCuda.test_mempool PASS > >> test_driver.py:258: TestCuda.test_mempool_2 PASS > >> test_driver.py:288: TestCuda.test_multi_context FAIL > >> test_driver.py:194: TestCuda.test_multichannel_2d_texture terminate > >> called after throwing an instance of 'cuda::error' > >> what(): cuMemFree failed: invalid context > >> Aborted > >> > >> > >> Any idea what's wrong? > >> > >> Best, > >> > >> N > >> > >> On Mon, Mar 16, 2009 at 2:56 AM, Philipp Lies < > >> > >> [email protected]> wrote: > >>> Hi, > >>> > >>> Andreas Klöckner wrote: > >>>> On Donnerstag 12 März 2009, Philipp Lies wrote: > >>>>> /opt/PyCUDA/pycuda-git/test$python test_driver.py > >>>>> ...E....Eterminate called after throwing an instance of 'cuda::error' > >>>>> what(): cuMemFree failed: invalid context > >>>>> Aborted > >>>> > >>>> Can you rerun this with '-v' so I can see which tests are affected? > >>> > >>> strange thing: it works now. I had to reboot in the meantime, so this > >>> seems to have solved the problem. Even though I worked with PyCUDA > >>> before the reboot. So I keep my eyes open and notify you when the > >>> problem reappears. > >>> Fixed tests work now, thx. > >>> > >>> Cheers > >>> > >>> Philipp > >>> > >>> > >>> _______________________________________________ > >>> PyCuda mailing list > >>> [email protected] > >>> http://tiker.net/mailman/listinfo/pycuda_tiker.net > >> > >> -- > >> Nicolas Pinto > >> Ph.D. Candidate, Brain & Computer Sciences > >> Massachusetts Institute of Technology, USA > >> http://web.mit.edu/pinto > > > > -- > > Nicolas Pinto > > Ph.D. Candidate, Brain & Computer Sciences > > Massachusetts Institute of Technology, USA > > http://web.mit.edu/pinto
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ PyCUDA mailing list [email protected] http://tiker.net/mailman/listinfo/pycuda_tiker.net
