The Compile tool is a C++ application that enables you to dump a loaded executable network blob. The tool is delivered as an executable file that can be run on both Linux* and Windows*. The tool is located in the <INSTALLROOT>/deployment_tools/inference_engine/lib/intel64/
directory on Linux and <INSTALL_DIR\deployment_tools\inference_engine\bin\intel64\Release>
on Windows.
The workflow of the Compile tool is as follows:
- Upon the start, the tool application reads command-line parameters and loads a network to the Inference Engine device.
- The application exports a blob with the compiled network and writes it to the output file.
Run the Compile Tool
Running the application with the -h
option yields the following usage message:
./compile_tool -h
Inference Engine:
API version ............ <version>
Build .................. <build>
compile_tool [OPTIONS]
[OPTIONS]:
-h Optional. Print the usage message.
-m <value> Required. Path to the XML model.
-d <value> Required. Target device name.
-o <value> Optional. Path to the output file. Default value: "<model_xml_file>.blob".
-c <value> Optional. Path to the configuration file. Default value: "config".
-ip <value> Optional. Specifies precision for all input layers of the network. Supported values: FP32, FP16, U8. Default value: FP32.
-op <value> Optional. Specifies precision for all output layers of the network. Supported values: FP32, FP16, U8. Default value: FP32.
-iop "<value>" Optional. Specifies precision for input and output layers by name.
By default, all inputs and outputs have the FP16 precision.
Available precisions: FP32, FP16, U8.
Example: -iop "input:FP16, output:FP16".
Notice that quotes are required.
Overwrites precision from ip and op options for specified layers.
VPU options:
-VPU_NUMBER_OF_SHAVES <value> Optional. Specifies number of shaves. Should be set with "VPU_NUMBER_OF_CMX_SLICES". Overwrites value from config.
-VPU_NUMBER_OF_CMX_SLICES <value> Optional. Specifies number of CMX slices. Should be set with "VPU_NUMBER_OF_SHAVES". Overwrites value from config.
-VPU_TILING_CMX_LIMIT_KB <value> Optional. Specifies CMX limit for data tiling in kB. Value should be equal or greater than -1, where -1 means default value of limit. Overwrites value from config.
DLA options:
-DLA_ARCH_NAME <value> Optional. Specify architecture name used to compile executable network for FPGA device.
Running the application with the empty list of options yields an error message.
To dump a blob using a trained Faster R-CNN network, use the command below:
./compile_tool -m <path_to_model>/model_name.xml
FPGA Option
You can compile executable network without a connected FPGA device with a loaded DLA bitstream. To do that, specify the architecture name of the DLA bitstream using the parameter -DLA_ARCH_NAME
.
Import and Export Functionality
Export
To save a blob file from your application, call the InferenceEngine::ExecutableNetwork::Export()
method:
std::ofstream file{"model_name.blob"}
executableNetwork.
Export(file);
Import
To import a blob with the network into your application, call the InferenceEngine::Core::ImportNetwork
method:
Example:
std::ifstream file{"model_name.blob"};
NOTE: Prior to the import, models must be converted to the Inference Engine format (*.xml + *.bin) using the Model Optimizer tool.