Ubuntu 16.04 CUDA8.0 CUDNN6.0 Caffe-gpu 配置記錄

最近由于遇到開(kāi)源代碼是基于Caffe的，所以打算在ubuntu平臺(tái)上配置caffe-gpu，本機(jī)硬件：
Ubuntu 16.04
GTX 1070 8g
CUDA 8.0 Cudnn 6
Anaconda Python3.6
Matlab R2017b

Let us go~

相關(guān)tips

#查看cudnn版本
cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2
or
cat /usr/include/cudnn.h | grep CUDNN_MAJOR -A 2

#查看cuda版本
cat /usr/local/cuda/version.txt
image.png

事實(shí)上這和官方的caffe版本是不同的，所以有些細(xì)節(jié)問(wèn)題bug可能在安裝官方caffe時(shí)不會(huì)碰到，anyway，舉一反三嘛

一、安裝依賴

For Ubuntu < 17.04:

#General dependencies
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install --no-install-recommends libboost-all-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev 
sudo apt-get install git cmake build-essential #以上兩條來(lái)自博客，是否必須有待確定(https://blog.csdn.net/yhaolpz/article/details/71375762）

BLAS：

sudo apt-get install libatlas-base-dev

二、編譯

2.1 編譯前的操作

主要是一些配置文件的改動(dòng):(https://blog.csdn.net/yhaolpz/article/details/71375762）

2.2 正式編譯

make all -j8

此步出錯(cuò):

CXX src/caffe/util/db_lmdb.cpp
CXX src/caffe/internal_thread.cpp
CXX src/caffe/util/signal_handler.cpp
CXX src/caffe/util/benchmark.cpp
PROTOC src/caffe/proto/caffe.proto
CXX src/caffe/layer_factory.cpp
CXX src/caffe/net.cpp
CXX src/caffe/util/im2col.cpp
CXX src/caffe/util/insert_splits.cpp
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/util/math_functions.hpp:9,
                 from src/caffe/util/im2col.cpp:4:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/util/math_functions.hpp:9,
                 from src/caffe/util/im2col.cpp:4:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/util/insert_splits.cpp:7:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/util/insert_splits.cpp:7:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/util/db.hpp:6,
                 from ./include/caffe/util/db_lmdb.hpp:10,
                 from src/caffe/util/db_lmdb.cpp:2:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/util/db.hpp:6,
                 from ./include/caffe/util/db_lmdb.hpp:10,
                 from src/caffe/util/db_lmdb.cpp:2:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/util/im2col.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/util/im2col.o] Error 1
make: *** Waiting for unfinished jobs....
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/util/db_lmdb.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/util/db_lmdb.o] Error 1
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/util/insert_splits.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/util/insert_splits.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/net.hpp:10,
                 from ./include/caffe/solver.hpp:7,
                 from ./include/caffe/util/signal_handler.h:5,
                 from src/caffe/util/signal_handler.cpp:7:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/net.hpp:10,
                 from ./include/caffe/solver.hpp:7,
                 from ./include/caffe/util/signal_handler.h:5,
                 from src/caffe/util/signal_handler.cpp:7:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/util/signal_handler.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/util/signal_handler.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/internal_thread.hpp:4,
                 from src/caffe/internal_thread.cpp:4:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/internal_thread.hpp:4,
                 from src/caffe/internal_thread.cpp:4:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/internal_thread.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/internal_thread.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/util/benchmark.cpp:3:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/util/benchmark.cpp:3:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/util/benchmark.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/util/benchmark.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/net.cpp:10:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/net.cpp:10:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/net.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/net.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layer.hpp:8,
                 from src/caffe/layer_factory.cpp:8:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:23:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:162:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:17:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layer.hpp:8,
                 from src/caffe/layer_factory.cpp:8:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetTanHLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:239:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetTanHLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:239:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetSoftmaxLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:216:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetSoftmaxLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:216:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetSigmoidLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:193:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetSigmoidLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:193:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetReLULayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:170:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetReLULayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:170:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetLRNLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:147:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetLRNLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:147:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetPoolingLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:110:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetPoolingLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:110:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetConvolutionLayer(const caffe::LayerParameter&) [with Dtype = double]’:
src/caffe/layer_factory.cpp:71:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
src/caffe/layer_factory.cpp: In function ‘boost::shared_ptr<caffe::Layer<Dtype> > caffe::GetConvolutionLayer(const caffe::LayerParameter&) [with Dtype = float]’:
src/caffe/layer_factory.cpp:71:1: warning: control reaches end of non-void function [-Wreturn-type]
 }
 ^
Makefile:583: recipe for target '/home/dlr/build_master_release/src/caffe/layer_factory.o' failed
make: *** [/home/dlr/build_master_release/src/caffe/layer_factory.o] Error 1

在github上找到解決放方法：

https://github.com/BVLC/caffe/issues/5793#issuecomment-404729152

翻過(guò)了第一個(gè)bug，下面有出錯(cuò)了：

CXX src/caffe/internal_thread.cpp
CXX src/caffe/util/im2col.cpp
CXX src/caffe/net.cpp
PROTOC src/caffe/proto/caffe.proto
CXX src/caffe/util/db_lmdb.cpp
CXX src/caffe/layer_factory.cpp
CXX src/caffe/util/benchmark.cpp
CXX src/caffe/util/signal_handler.cpp
CXX src/caffe/util/insert_splits.cpp
CXX src/caffe/util/db_leveldb.cpp
CXX src/caffe/util/cudnn.cpp
CXX src/caffe/util/blocking_queue.cpp
CXX src/caffe/util/db.cpp
CXX src/caffe/util/math_functions.cpp
CXX src/caffe/util/io.cpp
CXX src/caffe/util/upgrade_proto.cpp
CXX src/caffe/util/hdf5.cpp
CXX src/caffe/layers/prelu_layer.cpp
CXX src/caffe/layers/lrn_layer.cpp
CXX src/caffe/layers/eltwise_layer.cpp
CXX src/caffe/layers/softmax_layer.cpp
CXX src/caffe/layers/batch_reindex_layer.cpp
CXX src/caffe/layers/neuron_layer.cpp
CXX src/caffe/layers/image_data_layer.cpp
CXX src/caffe/layers/concat_layer.cpp
CXX src/caffe/layers/bnll_layer.cpp
CXX src/caffe/layers/cudnn_softmax_layer.cpp
CXX src/caffe/layers/power_layer.cpp
CXX src/caffe/layers/hinge_loss_layer.cpp
CXX src/caffe/layers/euclidean_loss_layer.cpp
CXX src/caffe/layers/slice_layer.cpp
CXX src/caffe/layers/cudnn_relu_layer.cpp
CXX src/caffe/layers/tile_layer.cpp
CXX src/caffe/layers/base_data_layer.cpp
CXX src/caffe/layers/hdf5_data_layer.cpp
src/caffe/layers/cudnn_relu_layer.cpp: In member function ‘virtual void caffe::CuDNNReLULayer<Dtype>::LayerSetUp(const std::vector<caffe::Blob<Dtype>*>&, const std::vector<caffe::Blob<Dtype>*>&)’:
src/caffe/layers/cudnn_relu_layer.cpp:14:3: error: ‘createTensorDesc’ is not a member of ‘caffe::cudnn’
   cudnn::createTensorDesc<Dtype>(&bottom_desc_);
   ^
src/caffe/layers/cudnn_relu_layer.cpp:14:32: error: expected primary-expression before ‘>’ token
   cudnn::createTensorDesc<Dtype>(&bottom_desc_);
                                ^
src/caffe/layers/cudnn_relu_layer.cpp:15:3: error: ‘createTensorDesc’ is not a member of ‘caffe::cudnn’
   cudnn::createTensorDesc<Dtype>(&top_desc_);
   ^
src/caffe/layers/cudnn_relu_layer.cpp:15:32: error: expected primary-expression before ‘>’ token
   cudnn::createTensorDesc<Dtype>(&top_desc_);
                                ^
src/caffe/layers/cudnn_relu_layer.cpp: In member function ‘virtual void caffe::CuDNNReLULayer<Dtype>::Reshape(const std::vector<caffe::Blob<Dtype>*>&, const std::vector<caffe::Blob<Dtype>*>&)’:
src/caffe/layers/cudnn_relu_layer.cpp:24:3: error: ‘setTensorNdDesc’ is not a member of ‘caffe::cudnn’
   cudnn::setTensorNdDesc<Dtype>(&bottom_desc_, bottom[0]->shape());
   ^
src/caffe/layers/cudnn_relu_layer.cpp:24:31: error: expected primary-expression before ‘>’ token
   cudnn::setTensorNdDesc<Dtype>(&bottom_desc_, bottom[0]->shape());
                               ^
src/caffe/layers/cudnn_relu_layer.cpp:25:3: error: ‘setTensorNdDesc’ is not a member of ‘caffe::cudnn’
   cudnn::setTensorNdDesc<Dtype>(&top_desc_, bottom[0]->shape());
   ^
src/caffe/layers/cudnn_relu_layer.cpp:25:31: error: expected primary-expression before ‘>’ token
   cudnn::setTensorNdDesc<Dtype>(&top_desc_, bottom[0]->shape());
                               ^
src/caffe/layers/cudnn_relu_layer.cpp: In instantiation of ‘void caffe::CuDNNReLULayer<Dtype>::Reshape(const std::vector<caffe::Blob<Dtype>*>&, const std::vector<caffe::Blob<Dtype>*>&) [with Dtype = float]’:
src/caffe/layers/cudnn_relu_layer.cpp:38:1:   required from here
src/caffe/layers/cudnn_relu_layer.cpp:24:46: warning: left operand of comma operator has no effect [-Wunused-value]
   cudnn::setTensorNdDesc<Dtype>(&bottom_desc_, bottom[0]->shape());
                                              ^
src/caffe/layers/cudnn_relu_layer.cpp:25:43: warning: left operand of comma operator has no effect [-Wunused-value]
   cudnn::setTensorNdDesc<Dtype>(&top_desc_, bottom[0]->shape());
                                           ^
src/caffe/layers/cudnn_relu_layer.cpp: In instantiation of ‘void caffe::CuDNNReLULayer<Dtype>::Reshape(const std::vector<caffe::Blob<Dtype>*>&, const std::vector<caffe::Blob<Dtype>*>&) [with Dtype = double]’:
src/caffe/layers/cudnn_relu_layer.cpp:38:1:   required from here
src/caffe/layers/cudnn_relu_layer.cpp:24:46: warning: left operand of comma operator has no effect [-Wunused-value]
   cudnn::setTensorNdDesc<Dtype>(&bottom_desc_, bottom[0]->shape());
                                              ^
src/caffe/layers/cudnn_relu_layer.cpp:25:43: warning: left operand of comma operator has no effect [-Wunused-value]
   cudnn::setTensorNdDesc<Dtype>(&top_desc_, bottom[0]->shape());
                                           ^
Makefile:583: recipe for target '/home/dlr/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/src/caffe/layers/cudnn_relu_layer.o' failed
make: *** [/home/dlr/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/src/caffe/layers/cudnn_relu_layer.o] Error 1
make: *** Waiting for unfinished jobs....

https://blog.csdn.net/qq_15505637/article/details/78423233

https://github.com/BVLC/caffe/issues/6139#issuecomment-407317661

注意的是SSCNet作者提到caffe
經(jīng)歷了兩天，終于到了最后的一步

https://github.com/muupan/dqn-in-the-caffe/issues/3#issuecomment-70795202

編譯時(shí) 避免多版本protobuf 沖突；

conda uninstall protobuf
conda uninstall libprotobuf

但注意在pycaffe編譯成功后再裝回來(lái)，否則import caffe會(huì)報(bào)錯(cuò)找不到google.protobuf

conda install libprotobuf

https://github.com/BVLC/caffe/issues/3046#issuecomment-316606472
(可能我這樣有些暴力。。)

接著可以開(kāi)始運(yùn)行demo了：

python demotest_model.py

然而又有新bug：


圖 1.png
E1031 20:10:34.877794 22173 common.cpp:113] Cannot create Cublas handle. Cublas won't be available.
F1031 20:10:35.117483 22163 cudnn_conv_layer.cpp:53] Check failed: status == CUDNN_STATUS_SUCCESS (4 vs. 0)  CUDNN_STATUS_INTERNAL_ERROR
*** Check failure stack trace: ***
Aborted (core dumped)
解決方法 注釋掉 Cudnn模式，從新編譯。。（雖然不治本吧）

解決了：https://devtalk.nvidia.com/default/topic/1007071/cuda-setup-and-installation/cuda-error-when-running-matrixmulcublas-sample-ubuntu-16-04/

sudo rm -rf .nv/

運(yùn)行demo時(shí)，遇到問(wèn)題

圖 2.png

參考：https://github.com/shurans/sscnet/issues/28#issue-278987264 （也是暴力的方法）
然后重新編譯

但是遇到了之前解決好的問(wèn)題：


圖 3.png

而且之前的解決方法無(wú)效了？？

export LD_LIBRARY_PATH=/home/dlr/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/lib:/home/dlr/Downloads/cuda8_5.1/lib64:~/anaconda3/lib:$LD_LIBRARY_PATH
export PYTHONPATH=/home/dlr/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/python:$PYTHONPATH

參考https://github.com/BVLC/caffe/issues/5170#issuecomment-271965791
對(duì)于python3，編譯caffe時(shí)要選擇對(duì)應(yīng)版本的protobuf，于是安裝protobuf-3.6.0
參考：http://blog.const.net.cn/a/19247.htm
https://github.com/protocolbuffers/protobuf/releases
重新編譯，編譯通過(guò)，但在make pycaffe出現(xiàn)新問(wèn)題：

圖 4.png

好吧，放棄py3.6，在anoconda py2.7虛擬環(huán)境重新來(lái)過(guò)：

git clone https://github.com/shurans/sscnet

經(jīng)過(guò)py2.7實(shí)驗(yàn)，發(fā)現(xiàn)還是卡到上個(gè)相同問(wèn)題，不過(guò)原因是不匹配的protobuf版本，我把安裝的protobuf3.6.0卸載了：

#在protobuf下載的包的位置
make clean
make uninstall

缺少 caffe_pb.h:
https://github.com/BVLC/caffe/issues/1761
兜兜轉(zhuǎn)轉(zhuǎn)，再次用同樣辦法解決圖4的問(wèn)題

打算直接注釋對(duì)應(yīng)blob.cpp的語(yǔ)句。重新編譯，然而：


圖 5

這個(gè)問(wèn)題一度讓我絕望，可是在看了
https://blog.csdn.net/e01528/article/details/80636867
后，激發(fā)了靈感，這個(gè)問(wèn)題在于：
在~/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/python中沒(méi)有caffe_pb2.py,經(jīng)過(guò)在caffe3d_suncg下搜索，位于
~/Project/sscnet/caffe_code/caffe3d_suncg/python/caffe/proto/caffe_pb2.py
應(yīng)該要保證~/Project/sscnet/caffe_code/caffe3d_suncg/build_master_release/python/caffe/proto/下有caffe_pb2.py，只要從上面的位置復(fù)制過(guò)來(lái)就成功了
還有注意

conda install protobuf==2.6.1

終于import caffe 成功！


圖 6
最后編輯于 ：
?著作權(quán)歸作者所有,轉(zhuǎn)載或內(nèi)容合作請(qǐng)聯(lián)系作者
【社區(qū)內(nèi)容提示】社區(qū)部分內(nèi)容疑似由AI輔助生成，瀏覽時(shí)請(qǐng)結(jié)合常識(shí)與多方信息審慎甄別。
平臺(tái)聲明：文章內(nèi)容（如有圖片或視頻亦包括在內(nèi)）由作者上傳并發(fā)布，文章內(nèi)容僅代表作者本人觀點(diǎn)，簡(jiǎn)書(shū)系信息發(fā)布平臺(tái)，僅提供信息存儲(chǔ)服務(wù)。

相關(guān)閱讀更多精彩內(nèi)容

友情鏈接更多精彩內(nèi)容