search

xw-hu / SINet

IEEE Transactions on Intelligent Transportation Systems (TITS), 2019
Other
80 stars 22 forks source link

Cuda8.0 + cudnn v6.0 support #1

Closed skywalker9096 closed 6 years ago

skywalker9096 commented 6 years ago

I use cudnn 6.0, but failed to make caffe.

I tried to use cudnn v4, but still failed:

NVCC src/caffe/layers/log_layer.cu
NVCC src/caffe/layers/mvn_layer.cu
src/caffe/layers/cudnn_conv_layer.cu(33): error: argument of type "cudnnAddMode_t" is incompatible with parameter of type "const void *"
          detected during instantiation of "void caffe::CuDNNConvolutionLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &, const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &) [with Dtype=float]"
(115): here

src/caffe/layers/cudnn_conv_layer.cu(33): error: argument of type "const void *" is incompatible with parameter of type "cudnnTensorDescriptor_t"
          detected during instantiation of "void caffe::CuDNNConvolutionLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &, const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &) [with Dtype=float]"
(115): here

src/caffe/layers/cudnn_conv_layer.cu(33): error: argument of type "const void *" is incompatible with parameter of type "cudnnTensorDescriptor_t"
          detected during instantiation of "void caffe::CuDNNConvolutionLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &, const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &) [with Dtype=float]"
(115): here

src/caffe/layers/cudnn_conv_layer.cu(33): error: too many arguments in function call
          detected during instantiation of "void caffe::CuDNNConvolutionLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &, const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &) [with Dtype=float]"
(115): here

src/caffe/layers/cudnn_conv_layer.cu(33): error: argument of type "cudnnAddMode_t" is incompatible with parameter of type "const void *"
          detected during instantiation of "void caffe::CuDNNConvolutionLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &, const std::vector<caffe::Blob<Dtype> *, std::allocator<caffe::Blob<Dtype> *>> &) [with Dtype=double]"
(115): here

is there updates for cudnn new versions? thx!

xw-hu commented 6 years ago

This code is implemented on CUDA 7.0 and cuDNN v3. You can compile the codes on other CUDA version without cuDNN.