src-d / kmcuda

Large scale K-means and K-nn implementation on NVIDIA GPU / CUDA
Other
791 stars 145 forks source link

Trying to install on Ubuntu 16.04 #1

Closed charlesgillespie closed 8 years ago

charlesgillespie commented 8 years ago

Hello, thanks for making a very cool package.

I am trying to install it on Ubuntu 16.04. I installed python3-dev and that resolved some of the python dependency issues but it is now failing with the following errors:

charles@srv-hanson-ml1:~/kmcuda-master$ cmake -DCMAKE_BUILD_TYPE=Release . && make
-- Configuring done
-- Generating done
-- Build files have been written to: /home/charles/kmcuda-master
[ 25%] Building CXX object CMakeFiles/KMCUDA.dir/python.cpp.o
/home/charles/kmcuda-master/python.cpp: In function ‘PyObject* py_kmeans_cuda(PyObject*, PyObject*, PyObject*)’:
/home/charles/kmcuda-master/python.cpp:109:53: error: invalid conversion from ‘int’ to ‘const float*’ [-fpermissive]
       verbosity, -1, samples, centroids, assignments);
                                                     ^
/home/charles/kmcuda-master/python.cpp:109:53: error: cannot convert ‘float*’ to ‘uint32_t* {aka unsigned int*}’ for argument ‘12’ to ‘int kmeans_cuda(bool, float, float, uint32_t, uint16_t, uint32_t, uint32_t, uint32_t, int32_t, const float*, float*, uint32_t*)’
CMakeFiles/KMCUDA.dir/build.make:378: recipe for target 'CMakeFiles/KMCUDA.dir/python.cpp.o' failed
make[2]: *** [CMakeFiles/KMCUDA.dir/python.cpp.o] Error 1
CMakeFiles/Makefile2:67: recipe for target 'CMakeFiles/KMCUDA.dir/all' failed
make[1]: *** [CMakeFiles/KMCUDA.dir/all] Error 2
Makefile:83: recipe for target 'all' failed
make: *** [all] Error 2

Any help appreciated!

vmarkovtsev commented 8 years ago

Oops, sorry, this is an artifact from the multi-gpu branch.