search

Xtra-Computing / thundersvm

ThunderSVM: A Fast SVM Library on GPUs and CPUs
Apache License 2.0
1.55k stars 215 forks source link

build error thunderSVM Cannot open include file: 'cub/cub.cuh' #269

Open radhaagr opened 9 months ago

radhaagr commented 9 months ago

FileNotFoundError Traceback (most recent call last) ~\AppData\Local\Temp/ipykernel_15480/1246353311.py in ----> 1 from thundersvm import SVC

c:\users\<>\appdata\local\programs\python\python37\lib\site-packages\thundersvm__init__.py in 8 """ 9 name = "thundersvm" ---> 10 from .thundersvm import *

c:\users\<>\appdata\local\programs\python\python37\lib\site-packages\thundersvm\thundersvm.py in 50 thundersvm = CDLL(lib_path) 51 else: ---> 52 raise FileNotFoundError("Please build the library first!") 53 SVM_TYPE = ['c_svc', 'nu_svc', 'one_class', 'epsilon_svr', 'nu_svr'] 54 KERNEL_TYPE = ['linear', 'polynomial', 'rbf', 'sigmoid', 'precomputed']

FileNotFoundError: Please build the library first!

radhaagr commented 9 months ago

I tried to build the thunderSVM with VS 2019 and Cuda 10.2, but the build has failed with below error - Command to get the solution file- cmake .. -DCMAKE_WINDOWS_EXPORT_ALL_SYMBOLS=TRUE -DBUILD_SHARED_LIBS=TRUE -DBUILD_TESTS=OFF -DUSE_CUDA=ON -G "Visual Studio 16 2019" when I try to build the thundersvm solution , it results in below error 2>C:/thundersvm/src/thundersvm/kernel/kernelmatrix_kernel.cu(12): fatal error C1083: Cannot open include file: 'cub/cub.cuh': No such file or directory

If I include path of cub directory in CMake file - include_directories(“C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.2/include/thrust/system/cuda/detail”) it resulted in a lot of undefined symbol errors.

device../iterator/arg_index_input_iterator.cuh(120): error : KeyValuePair is not a template 2> 2>C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include\thrust\system\cuda\detail\cub\device\dispatch../../agent/agent_reduce.cuh(65): error : identifier "BlockReduceAlgorithm" is undefined 2> 2>C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include\thrust\system\cuda\detail\cub\device\dispatch../../agent/agent_reduce.cuh(66): error : identifier "CacheLoadModifier" is undefined 2> 2>C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2\include\thrust\system\cuda\detail\cub\device\dispatch../../agent/agent_reduce.cuh(76): error : identifier "BlockReduceAlgorithm" is undefined