void cudnn_convolutional_setup(layer *l)
{
cudnnSetTensor4dDescriptor(l->dsrcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->ddstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetTensor4dDescriptor(l->srcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->dstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetTensor4dDescriptor(l->normTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, 1, l->out_c, 1, 1);
cudnnSetFilter4dDescriptor(l->dweightDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c/l->groups, l->size, l->size);
cudnnSetFilter4dDescriptor(l->weightDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c/l->groups, l->size, l->size);
#if CUDNN_MAJOR >= 6
cudnnSetConvolution2dDescriptor(l->convDesc, l->pad, l->pad, l->stride, l->stride, 1, 1, CUDNN_CROSS_CORRELATION, CUDNN_DATA_FLOAT);
#else
cudnnSetConvolution2dDescriptor(l->convDesc, l->pad, l->pad, l->stride, l->stride, 1, 1, CUDNN_CROSS_CORRELATION);
#endif
#if CUDNN_MAJOR >= 7
cudnnSetConvolutionGroupCount(l->convDesc, l->groups);
#else
if(l->groups > 1){
error("CUDNN < 7 doesn't support groups, please upgrade!");
}
#endif
cudnnGetConvolutionForwardAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->weightDesc,
l->convDesc,
l->dstTensorDesc,
CUDNN_CONVOLUTION_FWD_PREFER_FASTEST,
0,
&l->fw_algo);
cudnnGetConvolutionBackwardDataAlgorithm(cudnn_handle(),
l->weightDesc,
l->ddstTensorDesc,
l->convDesc,
l->dsrcTensorDesc,
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST,
0,
&l->bd_algo);
cudnnGetConvolutionBackwardFilterAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->ddstTensorDesc,
l->convDesc,
l->dweightDesc,
CUDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST,
0,
&l->bf_algo);
}
void cudnn_convolutional_setup(layer *l)
{
cudnnSetTensor4dDescriptor(l->dsrcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->ddstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetFilter4dDescriptor(l->dweightDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c, l->size, l->size);
cudnnSetTensor4dDescriptor(l->srcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->dstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetFilter4dDescriptor(l->weightDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c, l->size, l->size);
//cudnnSetConvolution2dDescriptor(l->convDesc, l->pad, l->pad, l->stride, l->stride, 1, 1, CUDNN_CROSS_CORRELATION);
cudnnSetConvolution2dDescriptor(l->convDesc, l->pad, l->pad, l->stride, l->stride, 1, 1, CUDNN_CROSS_CORRELATION, CUDNN_DATA_FLOAT);
cudnnGetConvolutionForwardAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->weightDesc,
l->convDesc,
l->dstTensorDesc,
CUDNN_CONVOLUTION_FWD_PREFER_FASTEST,
0,
&l->fw_algo);
cudnnGetConvolutionBackwardDataAlgorithm(cudnn_handle(),
l->weightDesc,
l->ddstTensorDesc,
l->convDesc,
l->dsrcTensorDesc,
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST,
0,
&l->bd_algo);
cudnnGetConvolutionBackwardFilterAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->ddstTensorDesc,
l->convDesc,
l->dweightDesc,
CUDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST,
0,
&l->bf_algo);
}
void CuDNNConvolutionLayer<Dtype>::Reshape(
const vector<Blob<Dtype>*>& bottom, const vector<Blob<Dtype>*>& top) {
ConvolutionLayer<Dtype>::Reshape(bottom, top);
CHECK_EQ(2, this->num_spatial_axes_)
<< "CuDNNConvolution input must have 2 spatial axes "
<< "(e.g., height and width). "
<< "Use 'engine: CAFFE' for general ND convolution.";
bottom_offset_ = this->bottom_dim_ / this->group_;
top_offset_ = this->top_dim_ / this->group_;
const int height = bottom[0]->shape(this->channel_axis_ + 1);
const int width = bottom[0]->shape(this->channel_axis_ + 2);
const int height_out = top[0]->shape(this->channel_axis_ + 1);
const int width_out = top[0]->shape(this->channel_axis_ + 2);
const int* pad_data = this->pad_.cpu_data();
const int pad_h = pad_data[0];
const int pad_w = pad_data[1];
const int* stride_data = this->stride_.cpu_data();
const int stride_h = stride_data[0];
const int stride_w = stride_data[1];
// Specify workspace limit for kernels directly until we have a
// planning strategy and a rewrite of Caffe's GPU memory mangagement
size_t workspace_limit_bytes, total_memory;
gpu_memory::getInfo(&workspace_limit_bytes, &total_memory);
for (int i = 0; i < bottom.size(); i++) {
cudnn::setTensor4dDesc<Dtype>(&bottom_descs_[i],
this->num_,
this->channels_ / this->group_, height, width,
this->channels_ * height * width,
height * width, width, 1);
cudnn::setTensor4dDesc<Dtype>(&top_descs_[i],
this->num_,
this->num_output_ / this->group_, height_out, width_out,
this->num_output_ * this->out_spatial_dim_,
this->out_spatial_dim_, width_out, 1);
cudnn::setConvolutionDesc<Dtype>(&conv_descs_[i], bottom_descs_[i],
filter_desc_, pad_h, pad_w, stride_h, stride_w);
if (!this->IsForwardPassed() || !this->IsBackwardPassed()) {
continue;
}
// choose forward and backward algorithms + workspace(s)
CUDNN_CHECK(cudnnGetConvolutionForwardAlgorithm(Caffe::cudnn_handle(),
bottom_descs_[i],
filter_desc_,
conv_descs_[i],
top_descs_[i],
CUDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes,
&fwd_algo_[i]));
CUDNN_CHECK(cudnnGetConvolutionForwardWorkspaceSize(Caffe::cudnn_handle(),
bottom_descs_[i],
filter_desc_,
conv_descs_[i],
top_descs_[i],
fwd_algo_[i],
&(workspace_fwd_sizes_[i])));
//
// choose backward algorithm for filter
CUDNN_CHECK(cudnnGetConvolutionBackwardFilterAlgorithm(
Caffe::cudnn_handle(),
bottom_descs_[i], top_descs_[i], conv_descs_[i], filter_desc_,
CUDNN_CONVOLUTION_BWD_FILTER_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes, &bwd_filter_algo_[i]) );
// get workspace for backwards filter algorithm
CUDNN_CHECK(cudnnGetConvolutionBackwardFilterWorkspaceSize(
Caffe::cudnn_handle(),
bottom_descs_[i], top_descs_[i], conv_descs_[i], filter_desc_,
bwd_filter_algo_[i], &workspace_bwd_filter_sizes_[i]));
// choose backward algo for data
CUDNN_CHECK(cudnnGetConvolutionBackwardDataAlgorithm(
Caffe::cudnn_handle(),
filter_desc_, top_descs_[i], conv_descs_[i], bottom_descs_[i],
CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes, &bwd_data_algo_[i]));
// get workspace size
CUDNN_CHECK(cudnnGetConvolutionBackwardDataWorkspaceSize(
Caffe::cudnn_handle(),
filter_desc_, top_descs_[i], conv_descs_[i], bottom_descs_[i],
bwd_data_algo_[i], &workspace_bwd_data_sizes_[i]) );
}
// Tensor descriptor for bias.
if (this->bias_term_) {
cudnn::setTensor4dDesc<Dtype>(&bias_desc_,
1, this->num_output_ / this->group_, 1, 1);
}
}
int
APPLY_SPECIFIC(conv_gi)(CudaNdarray *kerns, CudaNdarray *output,
CudaNdarray *im, cudnnConvolutionDescriptor_t desc,
float alpha, float beta, CudaNdarray **input) {
cudnnStatus_t err = CUDNN_STATUS_SUCCESS;
if (CudaNdarray_HOST_DIMS(im)[1] != CudaNdarray_HOST_DIMS(kerns)[1]) {
PyErr_SetString(PyExc_ValueError,
"GpuDnnConv images and kernel must have the same stack size\n");
return 1;
}
if (c_set_tensorNd(output, APPLY_SPECIFIC(output)) == -1)
return 1;
if (c_set_filterNd(kerns, APPLY_SPECIFIC(kerns)) == -1)
return 1;
int nb_dim = CudaNdarray_NDIM(output);
#ifdef CONV_INPLACE
Py_XDECREF(*input);
*input = im;
Py_INCREF(*input);
#else
if (CudaNdarray_prep_output(input, nb_dim, CudaNdarray_HOST_DIMS(im)) != 0)
return 1;
if (beta != 0.0 && CudaNdarray_CopyFromCudaNdarray(*input, im))
return 1;
#endif
if (c_set_tensorNd(*input, APPLY_SPECIFIC(input)) == -1)
return 1;
#if defined(CUDNN_VERSION) && CUDNN_VERSION >= 3000
{
size_t worksize;
void *workspace;
cudnnConvolutionBwdDataAlgo_t chosen_algo;
if (CHOOSE_ALGO)
{
// A new convolution implementation should be selected, based either on
// timing or heuristics, if in one of the two following cases :
// - The implementation should only be chosen during the first execution
// of an apply node and this is the first execution of the apply node.
// - The implementation should be chosen as often as necessary and the
// shapes of the inputs differ from the last time an implementation
// was chosen.
bool reuse_previous_algo;
if (CHOOSE_ALGO_ONCE)
{
// Only choose a new implementation of none has been chosen before.
reuse_previous_algo = APPLY_SPECIFIC(previous_algo_set);
}
else
{
// Reuse the previous implementation if the the kernels and the outputs
// have the same shapes as they had when the previous implementation
// was selected
bool same_shapes = true;
for (int i = 0; (i < nb_dim) && same_shapes; i++)
{
same_shapes &= (CudaNdarray_HOST_DIMS(kerns)[i] ==
APPLY_SPECIFIC(previous_kerns_shape)[i]);
same_shapes &= (CudaNdarray_HOST_DIMS(output)[i] ==
APPLY_SPECIFIC(previous_output_shape)[i]);
}
reuse_previous_algo = same_shapes;
}
// If the previously choosen implementation can't be reused, select a
// new one based on the shapes of the current inputs
if (!reuse_previous_algo)
{
// Obtain a convolution algorithm appropriate for the kernel and output
// shapes. Either by choosing one according to heuristics or by making
// CuDNN time every implementation and choose the best one.
if (CHOOSE_ALGO_TIME)
{
// Time the different implementations to choose the best one
int requestedCount = 1;
int count;
cudnnConvolutionBwdDataAlgoPerf_t choosen_algo_perf;
err = cudnnFindConvolutionBackwardDataAlgorithm(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
requestedCount,
&count,
&choosen_algo_perf);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error selecting convolution algo: "
"%s", cudnnGetErrorString(err));
return 1;
}
chosen_algo = choosen_algo_perf.algo;
}
else
{
// Choose the convolution implementation using heuristics based on the
// shapes of the inputs and the amount of memory available.
// Get the amount of available memory
size_t free = 0, total = 0;
cudaError_t err2 = cudaMemGetInfo(&free, &total);
if (err2 != cudaSuccess){
cudaGetLastError();
fprintf(stderr,
"Error when trying to find the memory information"
" on the GPU: %s\n", cudaGetErrorString(err2));
return 1;
}
// Use heuristics to choose the implementation
err = cudnnGetConvolutionBackwardDataAlgorithm(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT,
free,
&chosen_algo);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error selecting convolution algo: %s",
cudnnGetErrorString(err));
return 1;
}
}
// Store the shapes of the kernels and output as well as the chosen
// algorithm for future use.
APPLY_SPECIFIC(previous_bwd_d_algo) = chosen_algo;
for (int i = 0; i < nb_dim; i++)
{
APPLY_SPECIFIC(previous_kerns_shape)[i] =
CudaNdarray_HOST_DIMS(kerns)[i];
APPLY_SPECIFIC(previous_output_shape)[i] =
CudaNdarray_HOST_DIMS(output)[i];
}
}
else
{
// Reuse the previously chosen convlution implementation
chosen_algo = APPLY_SPECIFIC(previous_bwd_d_algo);
}
}
else
{
chosen_algo = CONV_ALGO;
}
// The FFT implementation (only in v3 and onward) does not support strides,
// 1x1 filters or inputs with a spatial dimension larger than 1024.
// If the chosen implementation is FFT, validate that it can be used
// on the current data and default on a safe implementation if it
// can't.
if (chosen_algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT && nb_dim == 4)
{
// Extract the properties of the convolution descriptor
int pad_h, pad_w, stride_v, stride_h, upscale_x, upscale_y;
cudnnConvolutionMode_t mode;
err = cudnnGetConvolution2dDescriptor(desc, &pad_h, &pad_w,
&stride_v, &stride_h,
&upscale_x, &upscale_y,
&mode);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error getting convolution properties: %s",
cudnnGetErrorString(err));
return 1;
}
// Extract the spatial size of the filters
int filter_h = CudaNdarray_HOST_DIMS(kerns)[3];
int filter_w = CudaNdarray_HOST_DIMS(kerns)[4];
// Extract the spatial size of the input
int input_h = CudaNdarray_HOST_DIMS(*input)[3];
int input_w = CudaNdarray_HOST_DIMS(*input)[4];
// Ensure that the selected implementation supports the requested
// convolution. Fall back to a safe implementation otherwise.
if (stride_v != 1 || stride_h != 1 || input_h > 1024 ||
input_w > 1024 || (filter_h == 1 && filter_w == 1))
{
chosen_algo = CUDNN_CONVOLUTION_BWD_DATA_ALGO_0;
}
}
// Infer required workspace size from the chosen implementation
err = cudnnGetConvolutionBackwardDataWorkspaceSize(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
chosen_algo,
&worksize);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error getting worksize: %s",
cudnnGetErrorString(err));
return 1;
}
// Allocate workspace for the convolution
workspace = get_work_mem(worksize);
if (workspace == NULL && worksize != 0)
return 1;
// Perform the convolution
err = cudnnConvolutionBackwardData_v3(
_handle,
(void *)&alpha,
APPLY_SPECIFIC(kerns), CudaNdarray_DEV_DATA(kerns),
APPLY_SPECIFIC(output), CudaNdarray_DEV_DATA(output),
desc,
chosen_algo,
workspace, worksize,
(void *)&beta,
APPLY_SPECIFIC(input), CudaNdarray_DEV_DATA(*input));
}
#else
err = cudnnConvolutionBackwardData(
_handle,
(void *)&alpha,
APPLY_SPECIFIC(kerns), CudaNdarray_DEV_DATA(kerns),
APPLY_SPECIFIC(output), CudaNdarray_DEV_DATA(output),
desc,
(void *)&beta,
APPLY_SPECIFIC(input), CudaNdarray_DEV_DATA(*input));
#endif
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "GpuDnnConvGradI: error doing operation: %s",
cudnnGetErrorString(err));
return 1;
}
return 0;
}
int
APPLY_SPECIFIC(conv_gi)(CudaNdarray *kerns, CudaNdarray *output,
CudaNdarray *im, cudnnConvolutionDescriptor_t desc,
float alpha, float beta, CudaNdarray **input) {
cudnnStatus_t err = CUDNN_STATUS_SUCCESS;
if (CudaNdarray_HOST_DIMS(im)[1] != CudaNdarray_HOST_DIMS(kerns)[1]) {
PyErr_SetString(PyExc_ValueError,
"GpuDnnConv images and kernel must have the same stack size\n");
return 1;
}
int nb_dim = CudaNdarray_NDIM(output);
#ifdef CONV_INPLACE
Py_XDECREF(*input);
*input = im;
Py_INCREF(*input);
#else
if (CudaNdarray_prep_output(input, nb_dim, CudaNdarray_HOST_DIMS(im)) != 0)
return 1;
if (beta != 0.0 && CudaNdarray_CopyFromCudaNdarray(*input, im))
return 1;
#endif
if (CudaNdarray_DIMS(im)[0] == 0 || CudaNdarray_DIMS(kerns)[0] == 0 || CudaNdarray_DIMS(kerns)[1] == 0) {
cudaError_t err2 = cudaMemset((*input)->devdata, 0,
CudaNdarray_SIZE(*input) * sizeof(real));
if (err2 != cudaSuccess) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConv grad wrt. inputs could not fill the output with zeros: %s",
cudaGetErrorString(err2));
return 1;
}
return 0;
}
if (c_set_tensorNd(output, APPLY_SPECIFIC(output)) == -1)
return 1;
if (c_set_filterNd(kerns, APPLY_SPECIFIC(kerns)) == -1)
return 1;
if (c_set_tensorNd(*input, APPLY_SPECIFIC(input)) == -1)
return 1;
int expected_output_dims[5] = {0};
err = cudnnGetConvolutionNdForwardOutputDim(desc, APPLY_SPECIFIC(input), APPLY_SPECIFIC(kerns),
nb_dim, expected_output_dims);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "error computing convolution output dim: %s",
cudnnGetErrorString(err));
return 1;
}
if (nb_dim == 4) {
if ((CudaNdarray_HOST_DIMS(output)[0] != expected_output_dims[0]) ||
(CudaNdarray_HOST_DIMS(output)[1] != expected_output_dims[1]) ||
(CudaNdarray_HOST_DIMS(output)[2] != expected_output_dims[2]) ||
(CudaNdarray_HOST_DIMS(output)[3] != expected_output_dims[3])) {
PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %ldx%ldx%ldx%ld"
" but received gradient with shape %ldx%ldx%ldx%ld",
(long int)expected_output_dims[0], (long int)expected_output_dims[1],
(long int)expected_output_dims[2], (long int)expected_output_dims[3],
(long int)CudaNdarray_HOST_DIMS(output)[0], (long int)CudaNdarray_HOST_DIMS(output)[1],
(long int)CudaNdarray_HOST_DIMS(output)[2], (long int)CudaNdarray_HOST_DIMS(output)[3]);
return 1;
}
} else if (nb_dim == 5) {
if ((CudaNdarray_HOST_DIMS(output)[0] != expected_output_dims[0]) ||
(CudaNdarray_HOST_DIMS(output)[1] != expected_output_dims[1]) ||
(CudaNdarray_HOST_DIMS(output)[2] != expected_output_dims[2]) ||
(CudaNdarray_HOST_DIMS(output)[3] != expected_output_dims[3]) ||
(CudaNdarray_HOST_DIMS(output)[4] != expected_output_dims[4])) {
PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %ldx%ldx%ldx%ldx%ld"
" but received gradient with shape %ldx%ldx%ldx%ldx%ld",
(long int)expected_output_dims[0], (long int)expected_output_dims[1],
(long int)expected_output_dims[2], (long int)expected_output_dims[3],
(long int)expected_output_dims[4],
(long int)CudaNdarray_HOST_DIMS(output)[0], (long int)CudaNdarray_HOST_DIMS(output)[1],
(long int)CudaNdarray_HOST_DIMS(output)[2], (long int)CudaNdarray_HOST_DIMS(output)[3],
(long int)CudaNdarray_HOST_DIMS(output)[4]);
return 1;
}
}
{
size_t worksize;
void *workspace;
cudnnConvolutionBwdDataAlgo_t chosen_algo;
if (CHOOSE_ALGO)
{
// A new convolution implementation should be selected, based either on
// timing or heuristics, if in one of the two following cases :
// - The implementation should only be chosen during the first execution
// of an apply node and this is the first execution of the apply node.
// - The implementation should be chosen as often as necessary and the
// shapes of the inputs differ from the last time an implementation
// was chosen.
bool reuse_previous_algo;
if (CHOOSE_ALGO_ONCE)
{
// Only choose a new implementation of none has been chosen before.
reuse_previous_algo = APPLY_SPECIFIC(previous_algo_set);
}
else
{
// Reuse the previous implementation if the the kernels and the outputs
// have the same shapes as they had when the previous implementation
// was selected
bool same_shapes = true;
for (int i = 0; (i < nb_dim) && same_shapes; i++)
{
same_shapes &= (CudaNdarray_HOST_DIMS(kerns)[i] ==
APPLY_SPECIFIC(previous_kerns_shape)[i]);
same_shapes &= (CudaNdarray_HOST_DIMS(output)[i] ==
APPLY_SPECIFIC(previous_output_shape)[i]);
}
reuse_previous_algo = same_shapes;
}
// If the previously choosen implementation can't be reused, select a
// new one based on the shapes of the current inputs
if (!reuse_previous_algo)
{
// Obtain a convolution algorithm appropriate for the kernel and output
// shapes. Either by choosing one according to heuristics or by making
// cuDNN time every implementation and choose the best one.
if (CHOOSE_ALGO_TIME)
{
// Time the different implementations to choose the best one
int requestedCount = 1;
int count;
cudnnConvolutionBwdDataAlgoPerf_t choosen_algo_perf;
err = cudnnFindConvolutionBackwardDataAlgorithm(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
requestedCount,
&count,
&choosen_algo_perf);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error selecting convolution algo: "
"%s", cudnnGetErrorString(err));
return 1;
}
chosen_algo = choosen_algo_perf.algo;
}
else
{
// Choose the convolution implementation using heuristics based on the
// shapes of the inputs and the amount of memory available.
// Get the amount of available memory
size_t free = 0, total = 0;
cudaError_t err2 = cudaMemGetInfo(&free, &total);
if (err2 != cudaSuccess){
cudaGetLastError();
fprintf(stderr,
"Error when trying to find the memory information"
" on the GPU: %s\n", cudaGetErrorString(err2));
return 1;
}
// Use heuristics to choose the implementation
err = cudnnGetConvolutionBackwardDataAlgorithm(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT,
free,
&chosen_algo);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error selecting convolution algo: %s",
cudnnGetErrorString(err));
return 1;
}
}
// Store the shapes of the kernels and output as well as the chosen
// algorithm for future use.
APPLY_SPECIFIC(previous_bwd_d_algo) = chosen_algo;
APPLY_SPECIFIC(previous_algo_set) = true;
for (int i = 0; i < nb_dim; i++)
{
APPLY_SPECIFIC(previous_kerns_shape)[i] =
CudaNdarray_HOST_DIMS(kerns)[i];
APPLY_SPECIFIC(previous_output_shape)[i] =
CudaNdarray_HOST_DIMS(output)[i];
}
}
else
{
// Reuse the previously chosen convlution implementation
chosen_algo = APPLY_SPECIFIC(previous_bwd_d_algo);
}
}
else
{
chosen_algo = CONV_ALGO;
}
if (0){
char * a;
switch(chosen_algo){
case CUDNN_CONVOLUTION_BWD_DATA_ALGO_0:
a = "implicit gemm (0)";
break;
case CUDNN_CONVOLUTION_BWD_DATA_ALGO_1:
a = "precomp gemm (1)";
break;
case CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT:
a = "fft (2)";
break;
case CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT_TILING:
a = "fft tiling (3)";
break;
#if CUDNN_VERSION > 5000
case CUDNN_CONVOLUTION_BWD_DATA_ALGO_WINOGRAD:
a = "winograd (4)";
break;
#endif
}
printf("GpuDNNConvGI: algo %s\n", a);
}
// The FFT implementation (only in V3 and onward) does not support strides,
// 1x1 filters or inputs with a spatial dimension larger than 1024.
// The tiled-FFT implementation (only in V4 onward) does not support
// strides.
// If the chosen implementation is FFT or tiled-FFT, validate that it can
// be used on the current data and default on a safe implementation if it
// can't.
// Following code is 2d-specific, but it is fine as FFT and tiled-FFT are
// defined only for 2d-filters
if ((chosen_algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT_TILING ||
chosen_algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT) && nb_dim == 4)
{
// Extract the properties of the convolution descriptor
int nd;
int pad[2];
int stride[2];
int upscale[2];
cudnnConvolutionMode_t mode;
cudnnDataType_t data_type;
err = cudnnGetConvolutionNdDescriptor(desc, 2, &nd, pad, stride,
upscale, &mode, &data_type);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error getting convolution properties: %s",
cudnnGetErrorString(err));
return 1;
}
// Extract the spatial size of the filters
int filter_h = CudaNdarray_HOST_DIMS(kerns)[2];
int filter_w = CudaNdarray_HOST_DIMS(kerns)[3];
// Extract the spatial size of the input
int input_h = CudaNdarray_HOST_DIMS(*input)[2];
int input_w = CudaNdarray_HOST_DIMS(*input)[3];
// Ensure that the selected implementation supports the requested
// convolution. Fall back to a safe implementation otherwise.
if (chosen_algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT)
{
if (stride[0] != 1 || stride[1] != 1 || input_h > 1024 ||
input_w > 1024 || (filter_h == 1 && filter_w == 1))
{
chosen_algo = CUDNN_CONVOLUTION_BWD_DATA_ALGO_0;
}
}
else
{
// chosen_algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT_TILING
if (stride[0] != 1 || stride[1] != 1)
{
chosen_algo = CUDNN_CONVOLUTION_BWD_DATA_ALGO_0;
}
}
}
// Infer required workspace size from the chosen implementation
err = cudnnGetConvolutionBackwardDataWorkspaceSize(_handle,
APPLY_SPECIFIC(kerns),
APPLY_SPECIFIC(output),
desc,
APPLY_SPECIFIC(input),
chosen_algo,
&worksize);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"GpuDnnConvGradI: error getting worksize: %s",
cudnnGetErrorString(err));
return 1;
}
// Allocate workspace for the convolution
workspace = get_work_mem(worksize);
if (workspace == NULL && worksize != 0)
return 1;
// Perform the convolution
err = cudnnConvolutionBackwardData(
_handle,
(void *)&alpha,
APPLY_SPECIFIC(kerns), CudaNdarray_DEV_DATA(kerns),
APPLY_SPECIFIC(output), CudaNdarray_DEV_DATA(output),
desc,
chosen_algo,
workspace, worksize,
(void *)&beta,
APPLY_SPECIFIC(input), CudaNdarray_DEV_DATA(*input));
}
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "GpuDnnConvGradI: error doing operation: %s",
cudnnGetErrorString(err));
return 1;
}
return 0;
}
void resize_convolutional_layer(convolutional_layer *l, int w, int h)
{
l->w = w;
l->h = h;
int out_w = convolutional_out_width(*l);
int out_h = convolutional_out_height(*l);
l->out_w = out_w;
l->out_h = out_h;
l->outputs = l->out_h * l->out_w * l->out_c;
l->inputs = l->w * l->h * l->c;
l->output = realloc(l->output,
l->batch*out_h * out_w * l->n*sizeof(float));
l->delta = realloc(l->delta,
l->batch*out_h * out_w * l->n*sizeof(float));
#ifdef GPU
cuda_free(l->delta_gpu);
cuda_free(l->output_gpu);
l->delta_gpu = cuda_make_array(l->delta, l->batch*out_h*out_w*l->n);
l->output_gpu = cuda_make_array(l->output, l->batch*out_h*out_w*l->n);
#ifdef CUDNN
cudnnSetTensor4dDescriptor(l->dsrcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->ddstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetFilter4dDescriptor(l->dfilterDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c, l->size, l->size);
cudnnSetTensor4dDescriptor(l->srcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->dstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);
cudnnSetFilter4dDescriptor(l->filterDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l->n, l->c, l->size, l->size);
int padding = l->pad ? l->size/2 : 0;
cudnnSetConvolution2dDescriptor(l->convDesc, padding, padding, l->stride, l->stride, 1, 1, CUDNN_CROSS_CORRELATION);
cudnnGetConvolutionForwardAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->filterDesc,
l->convDesc,
l->dstTensorDesc,
CUDNN_CONVOLUTION_FWD_PREFER_FASTEST,
0,
&l->fw_algo);
cudnnGetConvolutionBackwardDataAlgorithm(cudnn_handle(),
l->filterDesc,
l->ddstTensorDesc,
l->convDesc,
l->dsrcTensorDesc,
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST,
0,
&l->bd_algo);
cudnnGetConvolutionBackwardFilterAlgorithm(cudnn_handle(),
l->srcTensorDesc,
l->ddstTensorDesc,
l->convDesc,
l->dfilterDesc,
CUDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST,
0,
&l->bf_algo);
#endif
#endif
l->workspace_size = get_workspace_size(*l);
}
convolutional_layer make_convolutional_layer(int batch, int h, int w, int c, int n, int size, int stride, int pad, ACTIVATION activation, int batch_normalize, int binary, int xnor)
{
int i;
convolutional_layer l = {0};
l.type = CONVOLUTIONAL;
l.h = h;
l.w = w;
l.c = c;
l.n = n;
l.binary = binary;
l.xnor = xnor;
l.batch = batch;
l.stride = stride;
l.size = size;
l.pad = pad;
l.batch_normalize = batch_normalize;
l.filters = calloc(c*n*size*size, sizeof(float));
l.filter_updates = calloc(c*n*size*size, sizeof(float));
l.biases = calloc(n, sizeof(float));
l.bias_updates = calloc(n, sizeof(float));
// float scale = 1./sqrt(size*size*c);
float scale = sqrt(2./(size*size*c));
for(i = 0; i < c*n*size*size; ++i) l.filters[i] = scale*rand_uniform(-1, 1);
int out_h = convolutional_out_height(l);
int out_w = convolutional_out_width(l);
l.out_h = out_h;
l.out_w = out_w;
l.out_c = n;
l.outputs = l.out_h * l.out_w * l.out_c;
l.inputs = l.w * l.h * l.c;
l.output = calloc(l.batch*out_h * out_w * n, sizeof(float));
l.delta = calloc(l.batch*out_h * out_w * n, sizeof(float));
if(binary){
l.binary_filters = calloc(c*n*size*size, sizeof(float));
l.cfilters = calloc(c*n*size*size, sizeof(char));
l.scales = calloc(n, sizeof(float));
}
if(xnor){
l.binary_filters = calloc(c*n*size*size, sizeof(float));
l.binary_input = calloc(l.inputs*l.batch, sizeof(float));
}
if(batch_normalize){
l.scales = calloc(n, sizeof(float));
l.scale_updates = calloc(n, sizeof(float));
for(i = 0; i < n; ++i){
l.scales[i] = 1;
}
l.mean = calloc(n, sizeof(float));
l.variance = calloc(n, sizeof(float));
l.rolling_mean = calloc(n, sizeof(float));
l.rolling_variance = calloc(n, sizeof(float));
}
#ifdef GPU
l.filters_gpu = cuda_make_array(l.filters, c*n*size*size);
l.filter_updates_gpu = cuda_make_array(l.filter_updates, c*n*size*size);
l.biases_gpu = cuda_make_array(l.biases, n);
l.bias_updates_gpu = cuda_make_array(l.bias_updates, n);
l.scales_gpu = cuda_make_array(l.scales, n);
l.scale_updates_gpu = cuda_make_array(l.scale_updates, n);
l.delta_gpu = cuda_make_array(l.delta, l.batch*out_h*out_w*n);
l.output_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n);
if(binary){
l.binary_filters_gpu = cuda_make_array(l.filters, c*n*size*size);
}
if(xnor){
l.binary_filters_gpu = cuda_make_array(l.filters, c*n*size*size);
l.binary_input_gpu = cuda_make_array(0, l.inputs*l.batch);
}
if(batch_normalize){
l.mean_gpu = cuda_make_array(l.mean, n);
l.variance_gpu = cuda_make_array(l.variance, n);
l.rolling_mean_gpu = cuda_make_array(l.mean, n);
l.rolling_variance_gpu = cuda_make_array(l.variance, n);
l.mean_delta_gpu = cuda_make_array(l.mean, n);
l.variance_delta_gpu = cuda_make_array(l.variance, n);
l.x_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n);
l.x_norm_gpu = cuda_make_array(l.output, l.batch*out_h*out_w*n);
}
#ifdef CUDNN
cudnnCreateTensorDescriptor(&l.srcTensorDesc);
cudnnCreateTensorDescriptor(&l.dstTensorDesc);
cudnnCreateFilterDescriptor(&l.filterDesc);
cudnnCreateTensorDescriptor(&l.dsrcTensorDesc);
cudnnCreateTensorDescriptor(&l.ddstTensorDesc);
cudnnCreateFilterDescriptor(&l.dfilterDesc);
cudnnCreateConvolutionDescriptor(&l.convDesc);
cudnnSetTensor4dDescriptor(l.dsrcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l.batch, l.c, l.h, l.w);
cudnnSetTensor4dDescriptor(l.ddstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l.batch, l.out_c, l.out_h, l.out_w);
cudnnSetFilter4dDescriptor(l.dfilterDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l.n, l.c, l.size, l.size);
cudnnSetTensor4dDescriptor(l.srcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l.batch, l.c, l.h, l.w);
cudnnSetTensor4dDescriptor(l.dstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l.batch, l.out_c, l.out_h, l.out_w);
cudnnSetFilter4dDescriptor(l.filterDesc, CUDNN_DATA_FLOAT, CUDNN_TENSOR_NCHW, l.n, l.c, l.size, l.size);
int padding = l.pad ? l.size/2 : 0;
cudnnSetConvolution2dDescriptor(l.convDesc, padding, padding, l.stride, l.stride, 1, 1, CUDNN_CROSS_CORRELATION);
cudnnGetConvolutionForwardAlgorithm(cudnn_handle(),
l.srcTensorDesc,
l.filterDesc,
l.convDesc,
l.dstTensorDesc,
CUDNN_CONVOLUTION_FWD_PREFER_FASTEST,
0,
&l.fw_algo);
cudnnGetConvolutionBackwardDataAlgorithm(cudnn_handle(),
l.filterDesc,
l.ddstTensorDesc,
l.convDesc,
l.dsrcTensorDesc,
CUDNN_CONVOLUTION_BWD_DATA_PREFER_FASTEST,
0,
&l.bd_algo);
cudnnGetConvolutionBackwardFilterAlgorithm(cudnn_handle(),
l.srcTensorDesc,
l.ddstTensorDesc,
l.convDesc,
l.dfilterDesc,
CUDNN_CONVOLUTION_BWD_FILTER_PREFER_FASTEST,
0,
&l.bf_algo);
#endif
#endif
l.workspace_size = get_workspace_size(l);
l.activation = activation;
fprintf(stderr, "Convolutional Layer: %d x %d x %d image, %d filters -> %d x %d x %d image\n", h,w,c,n, out_h, out_w, n);
return l;
}
int
APPLY_SPECIFIC(conv_gi)(PyGpuArrayObject *kerns, PyGpuArrayObject *output,
PyGpuArrayObject *im,
cudnnConvolutionDescriptor_t desc,
double alpha, double beta, PyGpuArrayObject **input,
PyGpuContextObject *c) {
cudnnStatus_t err = CUDNN_STATUS_SUCCESS;
float af = alpha, bf = beta;
void *alpha_p;
void *beta_p;
if (PyGpuArray_DIMS(im)[1] != PyGpuArray_DIMS(kerns)[1]) {
PyErr_SetString(PyExc_ValueError, "images and kernel must have the same "
"stack size");
return 1;
}
if (c_set_tensorNd(output, APPLY_SPECIFIC(output)) == -1)
return 1;
if (c_set_filter(kerns, APPLY_SPECIFIC(kerns)) == -1)
return 1;
switch (im->ga.typecode) {
case GA_DOUBLE:
alpha_p = (void *)α
beta_p = (void *)β
break;
case GA_FLOAT:
case GA_HALF:
alpha_p = (void *)⁡
beta_p = (void *)&bf;
break;
default:
PyErr_SetString(PyExc_TypeError, "Unsupported type in convolution");
return 1;
}
#ifdef CONV_INPLACE
Py_XDECREF(*input);
*input = im;
Py_INCREF(*input);
#else
if (theano_prep_output(input, PyGpuArray_NDIM(im), PyGpuArray_DIMS(im),
im->ga.typecode, GA_C_ORDER, c) != 0)
return 1;
if (beta != 0.0 && pygpu_move(*input, im))
return 1;
#endif
if (c_set_tensorNd(*input, APPLY_SPECIFIC(input)) == -1)
return 1;
cudnnConvolutionBwdDataAlgo_t algo = CONV_ALGO;
cuda_enter(c->ctx);
#ifdef CHOOSE_ALGO
static int reuse_algo = 0;
static cudnnConvolutionBwdDataAlgo_t prev_algo = CONV_ALGO;
#ifndef CHOOSE_ONCE
static size_t prev_kern_dims[5] = {0};
static size_t prev_top_dims[5] = {0};
reuse_algo = 1;
for (unsigned int i = 0; i < PyGpuArray_NDIM(kerns); i++) {
reuse_algo = (reuse_algo &&
PyGpuArray_DIM(kerns, i) == prev_kern_dims[i]);
reuse_algo = (reuse_algo &&
PyGpuArray_DIM(output, i) == prev_top_dims[i]);
}
#endif
if (!reuse_algo) {
#ifdef CHOOSE_TIME
int count;
cudnnConvolutionBwdDataAlgoPerf_t choice;
err = cudnnFindConvolutionBackwardDataAlgorithm(
APPLY_SPECIFIC(_handle), APPLY_SPECIFIC(input), APPLY_SPECIFIC(output), desc,
APPLY_SPECIFIC(kerns), 1, &count, &choice);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "error selecting convolution algo: %s",
cudnnGetErrorString(err));
cuda_exit(c->ctx);
return 1;
}
algo = choice.algo;
#else
size_t free = 0, total = 0;
cudaError_t err2 = cudaMemGetInfo(&free, &total);
if (err2 != cudaSuccess) {
cudaGetLastError();
PyErr_Format(PyExc_RuntimeError, "Error when trying to find the memory "
"information on the GPU: %s\n", cudaGetErrorString(err2));
cuda_exit(c->ctx);
return 1;
}
err = cudnnGetConvolutionBackwardDataAlgorithm(
APPLY_SPECIFIC(_handle), APPLY_SPECIFIC(input), APPLY_SPECIFIC(output),
desc, APPLY_SPECIFIC(kerns),
CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT, free, &algo);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "error selecting convolution algo: %s",
cudnnGetErrorString(err));
cuda_exit(c->ctx);
return 1;
}
#endif
prev_algo = algo;
} else {
algo = prev_algo;
}
#ifdef CHOOSE_ONCE
reuse_algo = 1;
#else
for (unsigned int i = 0; i < PyGpuArray_NDIM(kerns); i++) {
prev_kern_dims[i] = PyGpuArray_DIM(kerns, i);
prev_top_dims[i] = PyGpuArray_DIM(output, i);
}
#endif
#endif
#if CUDNN_VERSION > 3000
if (algo == CUDNN_CONVOLUTION_BWD_DATA_ALGO_FFT) {
int nd;
int pad[2];
int stride[2];
int upscale[2];
cudnnConvolutionMode_t mode;
err = cudnnGetConvolutionNdDescriptor(desc, 2, &nd, pad, stride,
upscale, &mode);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError,
"error getting convolution properties: %s",
cudnnGetErrorString(err));
cuda_exit(c->ctx);
return 1;
}
if (stride[0] != 1 || stride[1] != 1 ||
PyGpuArray_DIM(*input, 2) > 1024 || PyGpuArray_DIM(*input, 3) > 1024 ||
(PyGpuArray_DIM(kerns, 2) == 1 && PyGpuArray_DIM(kerns, 3) == 1)) {
algo = CUDNN_CONVOLUTION_BWD_DATA_ALGO_0;
}
}
#endif
size_t worksize;
gpudata *workspace;
err = cudnnGetConvolutionBackwardDataWorkspaceSize(
APPLY_SPECIFIC(_handle), APPLY_SPECIFIC(kerns), APPLY_SPECIFIC(output), desc,
APPLY_SPECIFIC(input), algo, &worksize);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "error getting worksize: %s",
cudnnGetErrorString(err));
cuda_exit(c->ctx);
return 1;
}
if (worksize != 0) {
workspace = c->ops->buffer_alloc(c->ctx, worksize, NULL, 0, NULL);
if (workspace == NULL) {
PyErr_SetString(PyExc_RuntimeError,
"Could not allocate working memory");
cuda_exit(c->ctx);
return 1;
}
}
cuda_wait(kerns->ga.data, GPUARRAY_CUDA_WAIT_READ);
cuda_wait(output->ga.data, GPUARRAY_CUDA_WAIT_READ);
cuda_wait((*input)->ga.data, GPUARRAY_CUDA_WAIT_WRITE);
err = cudnnConvolutionBackwardData_v3(
APPLY_SPECIFIC(_handle),
alpha_p,
APPLY_SPECIFIC(kerns), PyGpuArray_DEV_DATA(kerns),
APPLY_SPECIFIC(output), PyGpuArray_DEV_DATA(output),
desc, algo, worksize == 0 ? NULL : *(void **)workspace, worksize,
beta_p,
APPLY_SPECIFIC(input), PyGpuArray_DEV_DATA(*input));
if (worksize != 0)
c->ops->buffer_release(workspace);
cuda_record(kerns->ga.data, GPUARRAY_CUDA_WAIT_READ);
cuda_record(output->ga.data, GPUARRAY_CUDA_WAIT_READ);
cuda_record((*input)->ga.data, GPUARRAY_CUDA_WAIT_WRITE);
cuda_exit(c->ctx);
if (err != CUDNN_STATUS_SUCCESS) {
PyErr_Format(PyExc_RuntimeError, "error doing operation: %s",
cudnnGetErrorString(err));
return 1;
}
return 0;
}
void CuDNNConvolutionLayer<Dtype>::Reshape(
const vector<Blob<Dtype>*>& bottom, const vector<Blob<Dtype>*>& top) {
ConvolutionLayer<Dtype>::Reshape(bottom, top);
CHECK_EQ(2, this->num_spatial_axes_)
<< "CuDNNConvolution input must have 2 spatial axes "
<< "(e.g., height and width). "
<< "Use 'engine: CAFFE' for general ND convolution.";
bottom_offset_ = this->bottom_dim_ / this->group_;
top_offset_ = this->top_dim_ / this->group_;
const int_tp height = bottom[0]->shape(this->channel_axis_ + 1);
const int_tp width = bottom[0]->shape(this->channel_axis_ + 2);
const int_tp height_out = top[0]->shape(this->channel_axis_ + 1);
const int_tp width_out = top[0]->shape(this->channel_axis_ + 2);
const int_tp* pad_data = this->pad_.cpu_data();
const int_tp pad_h = pad_data[0];
const int_tp pad_w = pad_data[1];
const int_tp* stride_data = this->stride_.cpu_data();
const int_tp stride_h = stride_data[0];
const int_tp stride_w = stride_data[1];
// Specify workspace limit for kernels directly until we have a
// planning strategy and a rewrite of Caffe's GPU memory mangagement
uint_tp workspace_limit_bytes = 8*1024*1024;
for (int_tp i = 0; i < bottom.size(); i++) {
cudnn::setTensor4dDesc<Dtype>(&bottom_descs_[i],
this->num_,
this->channels_ / this->group_, height, width,
this->channels_ * height * width,
height * width, width, 1);
cudnn::setTensor4dDesc<Dtype>(&top_descs_[i],
this->num_,
this->num_output_ / this->group_, height_out, width_out,
this->num_output_ * this->out_spatial_dim_,
this->out_spatial_dim_, width_out, 1);
cudnn::setConvolutionDesc<Dtype>(&conv_descs_[i], bottom_descs_[i],
filter_desc_, pad_h, pad_w,
stride_h, stride_w);
// choose forward and backward algorithms + workspace(s)
CUDNN_CHECK(cudnnGetConvolutionForwardAlgorithm(handle_[0],
bottom_descs_[i],
filter_desc_,
conv_descs_[i],
top_descs_[i],
CUDNN_CONVOLUTION_FWD_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes,
&fwd_algo_[i]));
CUDNN_CHECK(cudnnGetConvolutionForwardWorkspaceSize(handle_[0],
bottom_descs_[i],
filter_desc_,
conv_descs_[i],
top_descs_[i],
fwd_algo_[i],
&(workspace_fwd_sizes_[i])));
// choose backward algorithm for filter
CUDNN_CHECK(cudnnGetConvolutionBackwardFilterAlgorithm(handle_[0],
bottom_descs_[i], top_descs_[i], conv_descs_[i], filter_desc_,
CUDNN_CONVOLUTION_BWD_FILTER_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes, &bwd_filter_algo_[i]) );
// get workspace for backwards filter algorithm
CUDNN_CHECK(cudnnGetConvolutionBackwardFilterWorkspaceSize(handle_[0],
bottom_descs_[i], top_descs_[i], conv_descs_[i], filter_desc_,
bwd_filter_algo_[i], &workspace_bwd_filter_sizes_[i]));
// choose backward algo for data
CUDNN_CHECK(cudnnGetConvolutionBackwardDataAlgorithm(handle_[0],
filter_desc_, top_descs_[i], conv_descs_[i], bottom_descs_[i],
CUDNN_CONVOLUTION_BWD_DATA_SPECIFY_WORKSPACE_LIMIT,
workspace_limit_bytes, &bwd_data_algo_[i]));
// get workspace size
CUDNN_CHECK(cudnnGetConvolutionBackwardDataWorkspaceSize(handle_[0],
filter_desc_, top_descs_[i], conv_descs_[i], bottom_descs_[i],
bwd_data_algo_[i], &workspace_bwd_data_sizes_[i]) );
}
// reduce over all workspace sizes to get a maximum to allocate / reallocate
uint_tp total_workspace_fwd = 0;
uint_tp total_workspace_bwd_data = 0;
uint_tp total_workspace_bwd_filter = 0;
for (uint_tp i = 0; i < bottom.size(); i++) {
total_workspace_fwd = std::max(total_workspace_fwd,
workspace_fwd_sizes_[i]);
total_workspace_bwd_data = std::max(total_workspace_bwd_data,
workspace_bwd_data_sizes_[i]);
total_workspace_bwd_filter = std::max(total_workspace_bwd_filter,
workspace_bwd_filter_sizes_[i]);
}
// get max over all operations
uint_tp max_workspace = std::max(total_workspace_fwd,
total_workspace_bwd_data);
max_workspace = std::max(max_workspace, total_workspace_bwd_filter);
// ensure all groups have enough workspace
uint_tp total_max_workspace = max_workspace *
(this->group_ * CUDNN_STREAMS_PER_GROUP);
// this is the total amount of storage needed over all groups + streams
if (total_max_workspace > workspaceSizeInBytes) {
LOG(INFO) << "Reallocating workspace storage: " << total_max_workspace;
workspaceSizeInBytes = total_max_workspace;
// free the existing workspace and allocate a new (larger) one
cudaFree(this->workspaceData);
cudaError_t err = cudaMalloc(&(this->workspaceData), workspaceSizeInBytes);
if (err != cudaSuccess) {
// force zero memory path
for (int_tp i = 0; i < bottom.size(); i++) {
workspace_fwd_sizes_[i] = 0;
workspace_bwd_filter_sizes_[i] = 0;
workspace_bwd_data_sizes_[i] = 0;
fwd_algo_[i] = CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM;
bwd_filter_algo_[i] = CUDNN_CONVOLUTION_BWD_FILTER_ALGO_0;
bwd_data_algo_[i] = CUDNN_CONVOLUTION_BWD_DATA_ALGO_0;
}
// NULL out all workspace pointers
for (int_tp g = 0; g < (this->group_ * CUDNN_STREAMS_PER_GROUP); g++) {
workspace[g] = NULL;
}
// NULL out underlying data
workspaceData = NULL;
workspaceSizeInBytes = 0;
}
// if we succeed in the allocation, set pointer aliases for workspaces
for (int_tp g = 0; g < (this->group_ * CUDNN_STREAMS_PER_GROUP); g++) {
workspace[g] = reinterpret_cast<char *>(workspaceData) + g*max_workspace;
}
}
// Tensor descriptor for bias.
if (this->bias_term_) {
cudnn::setTensor4dDesc<Dtype>(&bias_desc_,
1, this->num_output_ / this->group_, 1, 1);
}
}