IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[GNA] Support 1d tensors (#4270)

Browse Source
This commit is contained in:
Elizaveta Lobanova
2021-02-12 16:26:10 +03:00
committed by GitHub
parent 3dbecf01fc
commit f4c071f635
4 changed files with 106 additions and 94 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
inference-engine/src/gna_plugin/gna_graph_compiler.cpp
View File

@@ -1283,8 +1283,9 @@ void GNAGraphCompiler::AffinePrimitive(InferenceEngine::CNNLayerPtr layer, bool
auto input_data = HasTo2DReshapeData(layer) ? Get2DReshapedData(inputs, 8) : inputs;
auto in_dims = input_data->getDims();
uint32_t num_rows_in = InferenceEngine::details::product(in_dims) / in_dims.front();
uint32_t num_columns_in = in_dims.front();
auto batch_size = (in_dims.size() == 1) ? 1 : in_dims.front();
uint32_t num_rows_in = InferenceEngine::details::product(in_dims) / batch_size;
uint32_t num_columns_in = batch_size;
uint32_t num_rows_out = isDiag ? num_rows_in : FROM_IR_DIM(outputs, 1);
uint32_t num_padding = ALIGN(num_rows_in, 8) - num_rows_in;
uint32_t num_padding_out = isDiag ? num_padding : 0;

3
inference-engine/src/gna_plugin/gna_groups.hpp
View File

@@ -31,7 +31,8 @@ inline InferenceEngine::DataPtr Get2DReshapedData(InferenceEngine::DataPtr input
}
}
InferenceEngine::SizeVector newDims(dims.size(), 1);
size_t newDimsSize = (dims.size() > 1) ? dims.size() : 2;
InferenceEngine::SizeVector newDims(newDimsSize, 1);
newDims[0] = numColumnsIn;
newDims[1] = numRowsIn;
return std::make_shared<InferenceEngine::Data>(input->getName(),

190
inference-engine/src/gna_plugin/gna_plugin.cpp
View File

@@ -1190,8 +1190,9 @@ uint32_t GNAPlugin::QueueInference(const InferenceEngine::BlobMap &inputs, Infer
int inputNum = 0;
for (auto &input : inputs) {
auto inputLayout = input.second->getTensorDesc().getLayout();
if (inputLayout != Layout::NC && inputLayout != Layout::CN && inputLayout != Layout::CHW && inputLayout != Layout::NCHW) {
THROW_GNA_EXCEPTION << "Expected input blob to have Layout::NC or Layout::CN, but was: "
if (inputLayout != Layout::C && inputLayout != Layout::NC && inputLayout != Layout::CN &&
inputLayout != Layout::CHW && inputLayout != Layout::NCHW) {
THROW_GNA_EXCEPTION << "Expected input blob to have Layout::C, Layout::NC, Layout::CN, Layout::NCHW or Layout::CHW. But was: "
<< input.second->getTensorDesc().getLayout();
}
@@ -1206,7 +1207,7 @@ uint32_t GNAPlugin::QueueInference(const InferenceEngine::BlobMap &inputs, Infer
inputLayout = Layout::NC;
}
auto is2D = input.second->getTensorDesc().getLayout() == Layout::NC || input.second->getTensorDesc().getLayout() == Layout::CN;
auto is1D = input.second->getTensorDesc().getLayout() == Layout::C;
auto is3D = input.second->getTensorDesc().getLayout() == Layout::CHW;
if (!inputsDesc->ptr_inputs_global_id.count(input.first)) {
@@ -1232,9 +1233,9 @@ uint32_t GNAPlugin::QueueInference(const InferenceEngine::BlobMap &inputs, Infer
}
}
auto importedElements = is2D ? dims[dims.size() - 1] : dims[dims.size() - 1] * dims[dims.size() - 2] * dims[dims.size() - 3];
auto importedFrames = is3D ? 1 : dims[0];
auto targetGroups = is2D ? dims[dims.size() - 2] : dims[0]; // TODO: no proper support for groups yet
auto importedElements = is1D ? dims[0] : details::product(++std::begin(dims), std::end(dims));
auto importedFrames = (is3D || is1D) ? 1 : dims[0];
auto targetGroups = is1D ? 1 : dims[0]; // TODO: no proper support for groups yet
auto importedElementSizeBytes = gnaFlags->sw_fp32 ? 4 : 2;
auto importedBytes = importedElements * importedFrames * importedElementSizeBytes;
@@ -1257,8 +1258,8 @@ uint32_t GNAPlugin::QueueInference(const InferenceEngine::BlobMap &inputs, Infer
auto transpose_info = transpose_inputs_info.find(input.first);
if (transpose_info != std::end(transpose_inputs_info)) {
size_t batchSize = dims[0];
size_t elementsPerBatch = InferenceEngine::details::product(dims) / dims[0];
size_t batchSize = (dims.size() > 1) ? dims[0] : 1;
size_t elementsPerBatch = (dims.size() > 1) ? InferenceEngine::details::product(dims) / dims[0] : dims[0];
size_t transposed_data_size = 0;
for (const auto &part_transposition_info : transpose_info->second) {
transposed_data_size += part_transposition_info.num_transpose_rows * part_transposition_info.num_transpose_columns;
@@ -1345,89 +1346,87 @@ GnaWaitStatus GNAPlugin::WaitFor(uint32_t request_idx, int64_t millisTimeout) {
for (auto && outputBlobIt : request) {
auto & outputBlob = outputBlobIt.second;
auto & outputDesc = outputsDesc[output_idx];
if (outputBlob->getTensorDesc().getLayout() == Layout::NC || outputBlob->getTensorDesc().getLayout() == Layout::CN
|| outputBlob->getTensorDesc().getLayout() == Layout::NCHW || outputBlob->getTensorDesc().getLayout() == Layout::CHW) {
auto dims = outputBlob->getTensorDesc().getDims();
auto is2D = outputBlob->getTensorDesc().getLayout() == Layout::NC || outputBlob->getTensorDesc().getLayout() == Layout::CN;
auto is3D = outputBlob->getTensorDesc().getLayout() == Layout::CHW;
auto& exportOutputDims = outputBlob->getTensorDesc().getDims();
auto batchSize = is3D ? 1 : exportOutputDims[0];
auto elementsPerBatch = is2D ? exportOutputDims[exportOutputDims.size() - 1]
: exportOutputDims[exportOutputDims.size() - 1]
* exportOutputDims[exportOutputDims.size() - 2]
* exportOutputDims[exportOutputDims.size() - 3];
auto transpose_output_info = transpose_outputs_info.find(outputBlobIt.first);
if (transpose_output_info != std::end(transpose_outputs_info)) {
size_t transposed_data_size = 0;
for (const auto &part_transposition_info : transpose_output_info->second) {
transposed_data_size += part_transposition_info.num_transpose_rows * part_transposition_info.num_transpose_columns;
}
if (elementsPerBatch != transposed_data_size) {
THROW_GNA_EXCEPTION << "Transposed data size (" << transposed_data_size
<< ") do not match output buffer length of " << elementsPerBatch;
}
TransposeTensorFromNCHWToNHWC(outputDesc.num_bytes_per_element,
batchSize,
elementsPerBatch,
reinterpret_cast<uint8_t*>(outputDesc.ptrs[request_idx]),
true,
transpose_output_info->second);
}
ExportScores(outputBlob->buffer(),
outputDesc.ptrs[request_idx],
outputDesc.orientation,
batchSize,
batchSize,
elementsPerBatch,
elementsPerBatch,
elementsPerBatch,
outputDesc.num_bytes_per_element,
sizeof(float));
if (gnadevice) {
#ifdef PLOT
FILE* f = nullptr;
static int num_infers = 0;
{
f = fopen("ex_scores.txt", "w");
}
num_infers++;
if (f) {
auto dims = outputBlob->getTensorDesc().getDims();
for (int i = 0; i < dims[dims.size() - 2]; i++) {
for (int j = 0; j < dims[dims.size() - 1]; j++) {
fprintf(f, "%d ", outputBlob->cbuffer().as<int32_t*>()[dims[dims.size() - 1] * i + j]);
}
fprintf(f, "\n");
}
fprintf(f, "\n\n");
}
#endif
ConvertToFloat(outputBlob->buffer(),
outputBlob->buffer(),
elementsPerBatch,
batchSize,
outputDesc.scale_factor);
#ifdef PLOT
if (f) {
auto dims = outputBlob->getTensorDesc().getDims();
for (int i = 0; i < dims[dims.size() - 2]; i++) {
for (int j = 0; j < dims[dims.size() - 1]; j++) {
fprintf(f, "%.2f ", outputBlob->cbuffer().as<float*>()[dims[dims.size() - 1] * i + j]);
}
fprintf(f, "\n");
}
fclose(f);
}
#endif
}
} else {
THROW_GNA_EXCEPTION << "Expected output blob to have Layout::NC, Layout::CN, Layout::NCHW or Layout::CHW. But was "
if (!outputBlob->getTensorDesc().getLayout() == Layout::C && !outputBlob->getTensorDesc().getLayout() == Layout::NC &&
!outputBlob->getTensorDesc().getLayout() == Layout::CN && !outputBlob->getTensorDesc().getLayout() == Layout::NCHW &&
!outputBlob->getTensorDesc().getLayout() == Layout::CHW) {
THROW_GNA_EXCEPTION << "Expected output blob to have Layout::C, Layout::NC, Layout::CN, Layout::NCHW or Layout::CHW. But was "
<< outputBlob->getTensorDesc().getLayout();
}
auto dims = outputBlob->getTensorDesc().getDims();
auto is1D = outputBlob->getTensorDesc().getLayout() == Layout::C;
auto is3D = outputBlob->getTensorDesc().getLayout() == Layout::CHW;
auto& exportOutputDims = outputBlob->getTensorDesc().getDims();
auto batchSize = (is1D || is3D) ? 1 : exportOutputDims[0];
auto elementsPerBatch = is1D ? exportOutputDims.front() :
details::product(++std::begin(exportOutputDims), std::end(exportOutputDims));
auto transpose_output_info = transpose_outputs_info.find(outputBlobIt.first);
if (transpose_output_info != std::end(transpose_outputs_info)) {
size_t transposed_data_size = 0;
for (const auto &part_transposition_info : transpose_output_info->second) {
transposed_data_size += part_transposition_info.num_transpose_rows * part_transposition_info.num_transpose_columns;
}
if (elementsPerBatch != transposed_data_size) {
THROW_GNA_EXCEPTION << "Transposed data size (" << transposed_data_size
<< ") do not match output buffer length of " << elementsPerBatch;
}
TransposeTensorFromNCHWToNHWC(outputDesc.num_bytes_per_element,
batchSize,
elementsPerBatch,
reinterpret_cast<uint8_t*>(outputDesc.ptrs[request_idx]),
true,
transpose_output_info->second);
}
ExportScores(outputBlob->buffer(),
outputDesc.ptrs[request_idx],
outputDesc.orientation,
batchSize,
batchSize,
elementsPerBatch,
elementsPerBatch,
elementsPerBatch,
outputDesc.num_bytes_per_element,
sizeof(float));
if (gnadevice) {
#ifdef PLOT
FILE* f = nullptr;
static int num_infers = 0;
{
f = fopen("ex_scores.txt", "w");
}
num_infers++;
if (f) {
for (int i = 0; i < batchSize; i++) {
for (int j = 0; j < dims[dims.size() - 1]; j++) {
fprintf(f, "%d ", outputBlob->cbuffer().as<int32_t*>()[dims[dims.size() - 1] * i + j]);
}
fprintf(f, "\n");
}
fprintf(f, "\n\n");
}
#endif
ConvertToFloat(outputBlob->buffer(),
outputBlob->buffer(),
elementsPerBatch,
batchSize,
outputDesc.scale_factor);
#ifdef PLOT
if (f) {
auto dims = outputBlob->getTensorDesc().getDims();
for (int i = 0; i < batchSize; i++) {
for (int j = 0; j < dims[dims.size() - 1]; j++) {
fprintf(f, "%.2f ", outputBlob->cbuffer().as<float*>()[dims[dims.size() - 1] * i + j]);
}
fprintf(f, "\n");
}
fclose(f);
}
#endif
}
output_idx++;
}
return GNA_REQUEST_COMPLETED;
@@ -1455,11 +1454,22 @@ bool GNAPlugin::Infer(const InferenceEngine::BlobMap &input, InferenceEngine::Bl
return Wait(QueueInference(input, result));
}
static InferenceEngine::Layout GetLayoutForDims(const InferenceEngine::SizeVector &dims) {
switch (dims.size()) {
case 1: return C;
case 2: return NC;
case 3: return CHW;
case 4: return NCHW;
default:
THROW_GNA_EXCEPTION << "Unsupported dimensions size in GNA: " << dims.size();
}
}
Blob::Ptr GNAPlugin::GetOutputBlob(const std::string& name, InferenceEngine::Precision precision) {
// need to have intermediate blob for interleave conversion
InferenceEngine::Blob::Ptr outputBlob;
auto outputDims = outputsDataMap[name]->getTensorDesc().getDims();
outputBlob = make_blob_with_precision(TensorDesc(precision, outputDims, outputDims.size() == 2 ? NC : (outputDims.size() == 3 ? CHW : NCHW)));
outputBlob = make_blob_with_precision(TensorDesc(precision, outputDims, GetLayoutForDims(outputDims)));
outputBlob->allocate();
return outputBlob;
}
@@ -1469,7 +1479,7 @@ Blob::Ptr GNAPlugin::GetInputBlob(const std::string& name, InferenceEngine::Prec
// need to have intermediate blob for interleave conversion
// TODO: NCHW format support is experimental = c++ MO did insert reshape, while TF mo - not
auto inputDims = inputsDataMap[name]->getTensorDesc().getDims();
inputBlob = make_blob_with_precision(TensorDesc(precision, inputDims, inputDims.size() == 2 ? NC : (inputDims.size() == 3 ? CHW : NCHW)));
inputBlob = make_blob_with_precision(TensorDesc(precision, inputDims, GetLayoutForDims(inputDims)));
inputBlob->allocate();
return inputBlob;
}

2
inference-engine/tests/unit/gna/gna_wait_test.cpp
View File

@@ -25,7 +25,7 @@ class GNAPluginForGNAWaitTest : public GNAPlugin {
public:
// Prepare underlining object to enable GNAInferRequest::Wait() working
GNAPluginForGNAWaitTest() {
InferenceEngine::TensorDesc td{ InferenceEngine::Precision::FP32, {}, InferenceEngine::Layout::HW };
InferenceEngine::TensorDesc td{ InferenceEngine::Precision::FP32, {1, 1}, InferenceEngine::Layout::HW };
auto fakeInfo = std::make_shared<InferenceEngine::InputInfo>();
auto fakePtr = std::make_shared<InferenceEngine::Data>("fakeName", td);
fakeInfo->setInputData(fakePtr);