kohr-h commented on issue #13135: [Python] CUDNN error from 3D deconvolution URL: https://github.com/apache/incubator-mxnet/issues/13135#issuecomment-436259702 More digging, again from `python -i tmp.py`: ```py >>> import ctypes, numpy as np >>> from mxnet.base import _LIB, py_str >>> self = output >>> data = np.ones(self.shape, dtype=self.dtype) # use ones to see when contents change >>> # First call fails >>> _LIB.MXNDArraySyncCopyToCPU(self.handle, data.ctypes.data_as(ctypes.c_void_p), ctypes.c_size_t(data.size)) -1 >>> py_str(_LIB.MXGetLastError()) '[14:49:18] c:\\users\\holger\\git\\mxnet\\src\\operator\\nn\\./cudnn/cudnn_deconvolution-inl.h:849: Failed to find any forward deconvolution algorithm with workspace size of 536870912 bytes, please consider reducing batch/model size or increasing the workspace size' >>> # Nothing changed in `data` >>> data.ravel()[:10] array([1., 1., 1., 1., 1., 1., 1., 1., 1., 1.], dtype=float32) >>> # Second call succeeds >>> _LIB.MXNDArraySyncCopyToCPU(self.handle, data.ctypes.data_as(ctypes.c_void_p), ctypes.c_size_t(data.size)) 0 >>> # Something changed, not clear if it makes sense >>> data.ravel()[:10] array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32) ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
