inline Dtype Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
Dtype loss = 0;
Reshape(bottom, top);
switch (Caffe::mode()) {
case Caffe::CPU:
Forward_cpu(bottom, top);
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->cpu_data();
const Dtype* loss_weights = top[top_id]->cpu_diff();
loss += caffe_cpu_dot(count, data, loss_weights);
}
break;
case Caffe::GPU:
Forward_gpu(bottom, top);
#ifndef CPU_ONLY
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->gpu_data();
const Dtype* loss_weights = top[top_id]->gpu_diff();
Dtype blob_loss = 0;
caffe_gpu_dot(count, data, loss_weights, &blob_loss);
loss += blob_loss;
}
#endif
break;
default:
LOG(FATAL) << "Unknown caffe mode.";
}
return loss;
}
inline Dtype Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
// Lock during forward to ensure sequential forward
Lock();
Dtype loss = 0;
Reshape(bottom, top);
switch (Caffe::mode()) {
case Caffe::CPU:
Forward_cpu(bottom, top);
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) {
continue;
}
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->cpu_data();
const Dtype* loss_weights = top[top_id]->cpu_diff();
loss += caffe_cpu_dot(count, data, loss_weights);
}
break;
case Caffe::GPU:
Forward_gpu(bottom, top);
#ifndef CPU_ONLY
if (device_->backend() == BACKEND_CUDA) {
#ifdef USE_CUDA
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) {
continue;
}
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->gpu_data();
const Dtype* loss_weights = top[top_id]->gpu_diff();
Dtype blob_loss = 0;
caffe_gpu_dot(count, data, loss_weights, &blob_loss);
loss += blob_loss;
}
#endif // USE_CUDA
} else {
#ifdef USE_GREENTEA
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) {
continue;
}
const int count = top[top_id]->count();
cl_mem data = (cl_mem) (top[top_id]->gpu_data());
cl_mem loss_weights = (cl_mem) (top[top_id]->gpu_diff());
Dtype blob_loss = 0;
greentea_gpu_dot(this->device_->id(), count, data, 0,
loss_weights, 0, &blob_loss);
loss += blob_loss;
}
#endif // USE_GREENTEA
}
#endif
break;
default:
LOG(FATAL)<< "Unknown caffe mode.";
}
Unlock();
return loss;
}
inline Dtype Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
// Lock during forward to ensure sequential forward
Lock();
Dtype loss = 0;
double tempdouble = 0;
int tempint = 0;
//if (bottom.size() != 0)
// tempint = bottom[0]->count();
Reshape(bottom, top);
switch (Caffe::mode()) {
case Caffe::CPU:
Forward_cpu(bottom, top);
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->cpu_data();
//const Dtype* label = bottom[0]->cpu_data();
//for (int i = 0; i < 10; i++)
//{
//
// printf("%lf\t", data[i]);
// /*if ((i + 1) % 10 == 0)
// {
// printf("\n");
// }*/
//
//}
//for (int i = 0; i < 10; i++)
// printf("%lf\t", label[i]);
const Dtype* loss_weights = top[top_id]->cpu_diff();
loss += caffe_cpu_dot(count, data, loss_weights);
}
break;
case Caffe::GPU:
Forward_gpu(bottom, top);
#ifndef CPU_ONLY
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->gpu_data();
const Dtype* loss_weights = top[top_id]->gpu_diff();
Dtype blob_loss = 0;
caffe_gpu_dot(count, data, loss_weights, &blob_loss);
loss += blob_loss;
}
#endif
break;
default:
LOG(FATAL) << "Unknown caffe mode.";
}
Unlock();
return loss;
}
inline void Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
vector<Blob<Dtype>*>* top) {
switch (Caffe::mode()) {
case Caffe::CPU:
Forward_cpu(bottom, top);
break;
case Caffe::GPU:
Forward_gpu(bottom, top);
break;
default:
LOG(FATAL)<< "Unknown caffe mode.";
}
};
inline Dtype Layer<Dtype>::Forward(const vector<Blob<Dtype>*>& bottom,
const vector<Blob<Dtype>*>& top) {
// Lock during forward to ensure sequential forward
for(int i=0; i<bottom.size(); i++)
{
LOG(INFO) << bottom[i]->shape_string();
}
Lock();
Dtype loss = 0;
Reshape(bottom, top);
switch (Caffe::mode()) {
/* case Caffe::CPU:
Forward_cpu(bottom, top);
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->cpu_data();
const Dtype* loss_weights = top[top_id]->cpu_diff();
loss += caffe_cpu_dot(count, data, loss_weights);
}
break;*/
case Caffe::GPU:
Forward_gpu(bottom, top);
#ifndef CPU_ONLY
for (int top_id = 0; top_id < top.size(); ++top_id) {
if (!this->loss(top_id)) { continue; }
const int count = top[top_id]->count();
const Dtype* data = top[top_id]->gpu_data();
const Dtype* loss_weights = top[top_id]->gpu_diff();
Dtype blob_loss = 0;
caffe_gpu_dot(count, data, loss_weights, &blob_loss);
loss += blob_loss;
}
#endif
break;
default:
LOG(FATAL) << "Unknown caffe mode.";
}
Unlock();
return loss;
}
void MaximumCaffe<T>::Forward(const std::vector<caffe::Blob<T>*>& bottom,
const std::vector<caffe::Blob<T>*>& top)
{
try
{
// CUDA
#ifdef USE_CUDA
Forward_gpu(bottom, top);
// OpenCL or CPU
#else
// CPU Version is already very fast (4ms)
Forward_cpu(bottom, top);
#endif
}
catch (const std::exception& e)
{
error(e.what(), __LINE__, __FUNCTION__, __FILE__);
}
}