ogmacorp / ogmaneodemos Goto Github PK

I have got this error by building the current version.
Please note I do not have any problem with the previous version...
Do you any idea?
Best thanks Binh
##############################
binh@ubuntu:~/RL/OgmaNeoDemos$ make
Scanning of target Runner
[ 5%] Building CXX object CMakeFiles/Runner.dir/demos/RunnerMain.cpp.o
[ 11%] Building CXX object CMakeFiles/Runner.dir/demos/runner/Runner.cpp.o
[ 17%] Linking CXX executable Runner
/usr/local/lib/libOgmaNeo.so: undefined reference to clEnqueueCopyImage@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clGetDeviceIDs@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseCommandQueue@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clRetainDevice@OPENCL_1.2'
/usr/local/lib/libOgmaNeo.so: undefined reference to clBuildProgram@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clCreateKernel@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseDevice@OPENCL_1.2' /usr/local/lib/libOgmaNeo.so: undefined reference to clGetPlatformIDs@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clSetKernelArg@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clEnqueueReadImage@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clEnqueueWriteImage@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clGetPlatformInfo@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clGetProgramInfo@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseProgram@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clCreateContext@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clEnqueueFillImage@OPENCL_1.2'
/usr/local/lib/libOgmaNeo.so: undefined reference to clCreateImage@OPENCL_1.2' /usr/local/lib/libOgmaNeo.so: undefined reference to clFinish@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clRetainKernel@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clGetProgramBuildInfo@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clRetainMemObject@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseContext@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clGetDeviceInfo@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clEnqueueNDRangeKernel@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clCreateProgramWithSource@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clGetImageInfo@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseKernel@OPENCL_1.0' /usr/local/lib/libOgmaNeo.so: undefined reference to clReleaseMemObject@OPENCL_1.0'
/usr/local/lib/libOgmaNeo.so: undefined reference to `clCreateCommandQueue@OPENCL_1.0'
collect2: error: ld returned 1 exit status
CMakeFiles/Runner.dir/build.make:127: recipe for target 'Runner' failed
make[2]: *** [Runner] Error 1
CMakeFiles/Makefile2:67: recipe for target 'CMakeFiles/Runner.dir/all' failed
make[1]: *** [CMakeFiles/Runner.dir/all] Error 2
Makefile:83: recipe for target 'all' failed
make: *** [all] Error 2

I understood that you use SparseFeatures(STDP, Chunk etc) for encodering the input images and then for predicting.
There is only one place for changing a type of encoder, using header file like
#input <neo/SparseFeaturesSTDP.h>
But I do not understand how this encoder is used internally.
Even by making a comment on this line, e.g.
//#input <neo/SparseFeaturesSTDP.h>
That means I do not use any SparseEncoder, but it does not matter anything. I can see what will be predicted.
I am not sure if the encoder is used correctly?

Do I understand your algorithm correctly?
Binh

building on a linux machine-

andrewcz@andrewcz-PORTEGE-Z30t-B ~/OgmaNeo/build $ cmake -DBUILD_SHARED_LIBS=ON ..
CMAKE_BUILD_TYPE not set, setting it to Release
-- Build type: Release
-- Bitness: 64
-- Shared libs: ON
-- Found existing OpenCL in OPENCL_INCLUDE_DIRS-NOTFOUND
-- Found existing OpenCL lib /usr/lib/x86_64-linux-gnu/libOpenCL.so
-- CL2 HPP not found!
-- Could NOT find flatbuffers (missing: FLATBUFFERS_FLATC_EXECUTABLE FLATBUFFERS_INCLUDE_DIR)
-- -- FlatBuffers will be downloaded and built from source
-- Found CL files - neoKernelsSparseFeaturesDistance.cl;neoKernelsSparseFeaturesChunk.cl;neoKernelsHierarchy.cl;neoKernelsPredictor.cl
CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
OPENCL_INCLUDE_DIRS
used as include directory in directory /home/andrewcz/OgmaNeo
used as include directory in directory /home/andrewcz/OgmaNeo
used as include directory in directory /home/andrewcz/OgmaNeo
used as include directory in directory /home/andrewcz/OgmaNeo
used as include directory in directory /home/andrewcz/OgmaNeo
used as include directory in directory /home/andrewcz/OgmaNeo

-- Configuring incomplete, errors occurred!
See also "/home/andrewcz/OgmaNeo/build/CMakeFiles/CMakeOutput.log".
andrewcz@andrewcz-PORTEGE-Z30t-B ~/OgmaNeo/build $

I have just downloaded the current version and have go the following error:
/home/binh/RL/OgmaNeoDemos/demos/Video_Prediction.cpp:19:27: fatal error: neo/Architect.h: No such file or directory
compilation terminated.
CMakeFiles/Video_Prediction.dir/build.make:62: recipe for target 'CMakeFiles/Video_Prediction.dir/demos/Video_Prediction.cpp.o' failed
make[2]: *** [CMakeFiles/Video_Prediction.dir/demos/Video_Prediction.cpp.o] Error 1
CMakeFiles/Makefile2:104: recipe for target 'CMakeFiles/Video_Prediction.dir/all' failed

Maybe after modifying the CMakeFiles, the include path fails ....

Thanks
Binh

Request: could you please provide a basic demo that does nothing fancy but just loads some numbers from file, does some learning, and spits out some numbers to the console. In other words, could you make a demo with only necessary dependencies. I like to study a Feynman machine without begin distracted by e.g. opencv, SFML etc.

The last two days I have been trying to get the two demo to work, and I failed.

Regards Henk-Jan.

@rcrowder Thank you for providing the sequence recall demo, much appreciated. As I was building my own demo I was wondering if an agent demo is available. Again, it does not need to be fancy, nor state of the art; working code would be enough to study the idea.

Btw, I'm really impressed with the small memory footprint and fast training of this particular method.