Register, Validate, and Deploy Custom Hyperbolic Tangent Layer Network to FPGA

This example shows how to register, validate, and deploy a custom hyperbolic tangent (tanh) layer network by using Deep Learning HDL Toolbox™. To deploy the network with the custom hyperbolic tangent (tanh) layer:

• Use the custom hyperbolic tangent (tanh) layer from Deep Learning Toolbox™.

• Register the custom hyperbolic tangent (tanh) layer by using the registerCustomLayer method.

• Validate the custom hyperbolic tangent (tanh) layer by generating a custom layer verification model.

• Generate a custom bitstream.

To retrieve the prediction results from the deployed custom layer network, use MATLAB®.

Create a Deep Learning Processor Configuration

To generate a custom processor configuration, use the dlhdl.ProcessorConfig object. The generated deep learning processor configuration object has a custom module that contains the preconfigured custom layers. Save the deep learning processor configuration to a variable hPC.

hPC = dlhdl.ProcessorConfig

hPC = 
                    Processing Module "conv"
                            ModuleGeneration: 'on'
                          LRNBlockGeneration: 'on'
                 SegmentationBlockGeneration: 'off'
                            ConvThreadNumber: 16
                             InputMemorySize: [227 227 3]
                            OutputMemorySize: [227 227 3]
                            FeatureSizeLimit: 2048

                      Processing Module "fc"
                            ModuleGeneration: 'on'
                      SoftmaxBlockGeneration: 'off'
                      SigmoidBlockGeneration: 'off'
                              FCThreadNumber: 4
                             InputMemorySize: 25088
                            OutputMemorySize: 4096

                  Processing Module "custom"
                            ModuleGeneration: 'on'
                              Multiplication: 'on'
                             InputMemorySize: 40
                            OutputMemorySize: 40

              Processor Top Level Properties
                              RunTimeControl: 'register'
                               RunTimeStatus: 'register'
                          InputStreamControl: 'register'
                         OutputStreamControl: 'register'
                           ProcessorDataType: 'single'

                     System Level Properties
                              TargetPlatform: 'Xilinx Zynq UltraScale+ MPSoC ZCU102 Evaluation Kit'
                             TargetFrequency: 200
                               SynthesisTool: 'Xilinx Vivado'
                             ReferenceDesign: 'AXI-Stream DDR Memory Access : 3-AXIM'
                     SynthesisToolChipFamily: 'Zynq UltraScale+'
                     SynthesisToolDeviceName: 'xczu9eg-ffvb1156-2-e'
                    SynthesisToolPackageName: ''
                     SynthesisToolSpeedValue: ''

Register Custom Layer and Model

To register an instance of the custom layer and custom layer Simulink® model use the registerCustomLayer method. This example leverages the built-in hyperbolic tangent (tanh) layer from Deep Learning Toolbox™. Deep Learning HDL Toolbox™ uses the Simulink® model to generate a verification model for the custom layer.

hTanhLayer = tanhLayer('Name','tanh')

hTanhLayer = 
  TanhLayer with properties:

    Name: 'tanh'

   Learnable Parameters
    No properties.

   State Parameters
    No properties.

  Show all properties

registerCustomLayer(hPC,Layer = hTanhLayer, Model = 'dnnfpgaTanhLayerModel.slx')

The custom deep learning processor configuration has a tanhLayer under the custom processing module. The custom hyperbolic tangent (tanh) layer is enabled by default for the bitstream generation.

Generate Verification Model for Custom Layer

Generate a verification model for your custom layer by using the openCustomLayerModel method. Generate a test network and a test image for your custom layer network by specifying blank arguments for the Network and InputImages arguments of the openCustomLayerModel method. The size of the test image matches the input layer size of the created test network.

openCustomLayerModel(hPC)

### The 'Network' property is empty for the given workflow object. An auto-generated network is provided.
### Notice: The layer 'input' with type 'nnet.cnn.layer.ImageInputLayer' is implemented in software.
### Notice: The layer 'output' with type 'nnet.cnn.layer.RegressionOutputLayer' is implemented in software.

### Allocating external memory buffers:

          offset_name          offset_address    allocated_space 
    _______________________    ______________    ________________

    "InputDataOffset"           "0x00000000"     "4.0 MB"        
    "OutputResultOffset"        "0x00400000"     "4.0 MB"        
    "SchedulerDataOffset"       "0x00800000"     "4.0 MB"        
    "SystemBufferOffset"        "0x00c00000"     "20.0 MB"       
    "InstructionDataOffset"     "0x02000000"     "4.0 MB"        
    "EndOffset"                 "0x02400000"     "Total: 36.0 MB"

The openCustomLayerModel method generates a verification model file called dnnfpgaCustomLayerVerificationModel.slx for your custom layer.

Simulate and Validate Custom Layer Model

Before you verify your custom layer model by using the verifyCustomLayerModel method, open the dnnfpgaCustomLayerVerificationModel.slx verification model. The verifyCustomLayerModel verifies the functionality of the custom layer and prediction accuracy of the network which has the custom layer.

verifyCustomLayerModel(hPC)

### Compiling Simulink model 'dnnfpgaCustomLayersTB' ...

Warning: '<a href="matlab:slprivate('open_and_hilite_port_hyperlink', 'hilite', ['dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler'], 'Inport', 17);">Input Port 17 (Right:8)</a>' of '<a href="matlab:open_and_hilite_hyperlink ('dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler','error')">dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler</a>' is not connected.

Warning: '<a href="matlab:slprivate('open_and_hilite_port_hyperlink', 'hilite', ['dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler'], 'Inport', 18);">Input Port 18 (Right:9)</a>' of '<a href="matlab:open_and_hilite_hyperlink ('dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler','error')">dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler</a>' is not connected.

Warning: '<a href="matlab:slprivate('open_and_hilite_port_hyperlink', 'hilite', ['dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler'], 'Inport', 19);">Input Port 19 (Right:10)</a>' of '<a href="matlab:open_and_hilite_hyperlink ('dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler','error')">dnnfpgaCustomLayersTB/DUT/Subsystem/Scheduler</a>' is not connected.

### Complete Simulink model 'dnnfpgaCustomLayersTB' compilation.
Verification passed.

Generate Custom Bitstream

Generate a custom bitstream that has the name myCustomLayer.bit by using the dlhdl.buildProcessor function. Save the generated bitstream to the myCustomLayer_prj folder.

dlhdl.buildProcessor(hPC,ProjectFolder = 'myCustomLayer_prj',ProcessorName ='myCustomLayer');

Deploy and Predict Custom Layer Network on Hardware

Deploy the custom layer network by creating a dlhdl.Workflow object with the custom network as the Network argument and the custom bitstream myCustomLayer.bit as the Bitstream argument. Compile and deploy the dlhdl.Workflow object. To retrieve the prediction results from the deployed network, use MATLAB® and the predict method.

layers = [imageInputLayer([2 2 4],'Name','data','Normalization', 'none');
tanhLayer('Name','tanh');
regressionLayer('Name','output');
]
myCustomNet = assembleNetwork(layers);
hTarget = dlhdl.Target('Xilinx','Interface','JTAG');
hW = dlhdl.Workflow(Network = myCustomNet, Bitstream = 'myCustomLayer.bit',Target = hTarget);
hW.compile;
hW.deploy;
image = randi(255, [2,2,4]);
hW.predict(single(image),Profile ='on');