kunrenzhilu
I use the follow code: ```python mat = np.repeat(np.repeat(np.arange(9).reshape(3,3), 3, axis=0), 3, axis=1).reshape(1,1,9,9) mat2 = torch.from_numpy(mat) P.im2col(Variable(mat2.cuda()), (3,3), (3,3), (0,0)) ``` Then it throws error: ``` --------------------------------------------------------------------------- CompileException Traceback (most recent call last) <ipython-input-29-8652fb792000> in <module>() ----> 1 P.im2col(Variable(mat2.cuda()), (3,3), (3,3), (0,0)) /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/pyinn/im2col.py in im2col(input, kernel_size, stride, padding) 244 TODO: add CPU version (via numpy?) 245 """ --> 246 return Im2Col(kernel_size, stride, padding)(input) 247 248 /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/pyinn/im2col.py in forward(self, input) 207 assert(input.is_contiguous()) 208 self.input_size = input.size()[-2:] --> 209 return im2col_batch(input, self.kernel_size, self.stride, self.padding) 210 211 def backward(self, grad_output): /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/pyinn/im2col.py in im2col_batch(input, kernel_size, stride, padding) 178 def im2col_batch(input, kernel_size, stride, padding): 179 if input.dim() == 3: --> 180 return _im2col(input, kernel_size, stride, padding) 181 elif input.dim() == 4: 182 shape = (input.size(0),) + im2col_shape(input.size()[1:], kernel_size, stride, padding) /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/pyinn/im2col.py in _im2col(data, kernel_size, stride, padding, out) 118 pad_h=pad_h, pad_w=pad_w, 119 stride_h=stride_h, stride_w=stride_w, --> 120 channels=n_input_plane) 121 f(block=(CUDA_NUM_THREADS,1,1), 122 grid=(GET_BLOCKS(n),1,1), cupy/util.pyx in cupy.util.memoize.decorator.ret() /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/pyinn/utils.pyc in load_kernel(kernel_name, code, **kwargs) 18 def load_kernel(kernel_name, code, **kwargs): 19 code = Template(code).substitute(**kwargs) ---> 20 kernel_code = cupy.cuda.compile_with_cache(code) 21 return kernel_code.get_function(kernel_name) /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/cupy/cuda/compiler.pyc in compile_with_cache(source, options, arch, cache_dir, extra_source) 150 return mod 151 --> 152 ptx = compile_using_nvrtc(source, options, arch) 153 ls = function.LinkState() 154 ls.add_ptr_data(ptx, six.u('cupy.ptx')) /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/cupy/cuda/compiler.pyc in compile_using_nvrtc(source, options, arch) 68 prog = _NVRTCProgram(source, cu_path) 69 try: ---> 70 ptx = prog.compile(options) 71 except CompileException as e: 72 dump = _get_bool_env_variable( /home/haibin2/data/pytorch_env/local/lib/python2.7/site-packages/cupy/cuda/compiler.pyc in compile(self, options) 231 except nvrtc.NVRTCError: 232 log = nvrtc.getProgramLog(self.ptr) --> 233 raise CompileException(log, self.src, self.name, options) 234 235 CompileException: /tmp/tmparMI_j/kern.cu(7): error: identifier "None" is undefined /tmp/tmparMI_j/kern.cu(7): error: identifier "None" is undefined 2 errors detected in the compilation of "/tmp/tmparMI_j/kern.cu". ``` I don't understand what is going wrong, do you have any idea?