Check failed: error == cudaSuccess (74 vs. 0) misaligned address

[wendalegood]

when train on Maxwell GeForce GTX TITAN X , there was an error( [math_functions.cu:79] Check failed: error == cudaSuccess (74 vs. 0) misaligned address) , I change cuda7.5 to cuda8.0 and cudnn to cudnn5.1 but it was not work. Then I change the convolution engine to CAFFE, it worked but the training speed was too slow. How to fix the problem, thx.

[shicai]

I have not met this error before, maybe there is something wrong with your caffe or cudnn? you may try https://github.com/yonghenglh6/DepthwiseConvolution for speedup.

[yytzjgsu]

yeah，I also met misaligned address problem and have no idea. What's shocking me more is the GPU memory it costs,almost 11G. Even training with a Titan X ,the batchsize could only be 1.By the way , the input size is 512x320。

[Ai-is-light]

I also meet this problem . Did you deal with it? @wendalegood @shicai @ @ @