Octave and CUDA

[mpender]

I'm trying to work through an example of building an oct file that links to
the CUDA blas library that I found in the paper "Technical Report on Matrix
Multiplication on GPU in Octave"
(http://docs.ihpc.a-star.edu.sg/papers/MMGPUOctave.pdf) and running into a
compiler error that I don't really understand.  The first program appears to
compile properly using nvcc, but the mkoctfile stage chokes.

Is 'rep' some kind of private data member that is not supposed to be
accessible?  I'm not really sure from the name 'rep' what it means.

address@hidden:~/octavework/test17$ nvcc run_sgemm.cu -c -shared -Xcompiler
-fPIC
address@hidden:~/octavework/test17$ mkoctfile gpu_sgemm.cc run_sgemm.o
-L/usr/local/cuda/lib -I/usr/local/cuda/include -lcuda -lcudart -lcublas
/usr/include/octave-3.2.4/octave/Array.h: In function ‘octave_value_list
Fgpu_sgemm(const octave_value_list&, int)’:
/usr/include/octave-3.2.4/octave/Array.h:110:32: error:
‘Array<double>::ArrayRep* Array<double>::rep’ is protected
gpu_sgemm.cc:94:32: error: within this context
/usr/include/octave-3.2.4/octave/Array.h:110:32: error:
‘Array<double>::ArrayRep* Array<double>::rep’ is protected
gpu_sgemm.cc:95:32: error: within this context
/usr/include/octave-3.2.4/octave/Array.h:110:32: error:
‘Array<double>::ArrayRep* Array<double>::rep’ is protected
gpu_sgemm.cc:96:32: error: within this context
address@hidden:~/octavework/test17$




--
View this message in context: 
http://octave.1599824.n4.nabble.com/Octave-and-CUDA-tp4660998.html
Sent from the Octave - General mailing list archive at Nabble.com.