Issue with NN IP Connection in Vivado After Bitfile Generation

[AbuAli3]

Hello HLS4ML Community,

I recently generated a bitfile for my design using HLS4ML and opened the block design in Xilinx Vivado. I noticed that all IPs are connected correctly, except for my Neural Network IP (myproject_axi_0).

Here are the steps I've taken:

Generated the bitfile successfully.
Opened the block design in Vivado

project_1.xpr

Run connection automation did not pop up, to automate myproject_axi_0.
Details:

Vivado Version: 2020.1
HLS4ML Version: 0.8.1

***** C/RTL SYNTHESIS *****
INFO: [SCHED 204-61] Option 'relax_ii_for_timing' is enabled, will increase II to preserve clock frequency constraints.
INFO: [HLS 200-10] Analyzing design file 'firmware/myproject.cpp' ...
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:43:71
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:43:75
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:47:74
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:47:79
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:55:67
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:55:71
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:59:74
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:59:79
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:67:67
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:67:71
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:69:76
WARNING: [HLS 214-113] Either use an argument of the function or declare the variable inside the dataflow loop body: firmware/myproject.cpp:69:81
WARNING: [HLS 200-471] Dataflow form checks found 12 issue(s) in file firmware/myproject.cpp
INFO: [HLS 200-10] Analyzing design file 'firmware/myproject_axi.cpp' ...
WARNING: [HLS 214-114] Since the only kind of statements allowed in a canonical dataflow region are variable declarations and function calls, the compiler may not be able to correctly handle the region: firmware/myproject_axi.cpp:17:2
WARNING: [HLS 214-114] Since the only kind of statements allowed in a canonical dataflow region are variable declarations and function calls, the compiler may not be able to correctly handle the region: firmware/myproject_axi.cpp:29:5
WARNING: [HLS 200-471] Dataflow form checks found 2 issue(s) in file firmware/myproject_axi.cpp
WARNING: [XFORM 203-631] Renaming function 'nnet::relu<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 32u>, nnet::array<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, 32u>, relu_config4>' to 'relu<array<ap_fixed,32u>,array<ap_ufixed<4,0,4,0,0>,32u>,relu_config4>' (firmware/nnet_utils/nnet_activation_stream.h:41:31)
WARNING: [XFORM 203-631] Renaming function 'nnet::relu<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 16u>, nnet::array<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, 16u>, relu_config7>' to 'relu<array<ap_fixed,16u>,array<ap_ufixed<4,0,4,0,0>,16u>,relu_config7>' (firmware/nnet_utils/nnet_activation_stream.h:41:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::normalize<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 32u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 32u>, config10>' to 'normalize<array<ap_fixed,32u>,array<ap_fixed<16,6,5,3,0>,32u>,config10>' (firmware/nnet_utils/nnet_batchnorm_stream.h:22:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::normalize<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 1u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 1u>, config12>' to 'normalize<array<ap_fixed,1u>,array<ap_fixed<16,6,5,3,0>,1u>,config12>' (firmware/nnet_utils/nnet_batchnorm_stream.h:22:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::normalize<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 16u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 16u>, config11>' to 'normalize<array<ap_fixed,16u>,array<ap_fixed<16,6,5,3,0>,16u>,config11>' (firmware/nnet_utils/nnet_batchnorm_stream.h:22:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense_wrapper<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, config8>' to 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config8>' (firmware/nnet_utils/nnet_dense_stream.h:13)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense_wrapper<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, config5>' to 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config5>' (firmware/nnet_utils/nnet_dense_stream.h:13)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense_wrapper<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, config2>' to 'dense_wrapper<ap_fixed<16, 6, 5, 3, 0>, ap_fixed<16, 6, 5, 3, 0>, config2>' (firmware/nnet_utils/nnet_dense_stream.h:13)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense<nnet::array<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, 32u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 16u>, config5>' to 'dense<array<ap_ufixed,32u>,array<ap_fixed<16,6,5,3,0>,16u>,config5>' (firmware/nnet_utils/nnet_dense_stream.h:36:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense<nnet::array<ap_ufixed<4, 0, (ap_q_mode)4, (ap_o_mode)0, 0>, 16u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 1u>, config8>' to 'dense<array<ap_ufixed,16u>,array<ap_fixed<16,6,5,3,0>,1u>,config8>' (firmware/nnet_utils/nnet_dense_stream.h:36:1)
WARNING: [XFORM 203-631] Renaming function 'nnet::dense<nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 50u>, nnet::array<ap_fixed<16, 6, (ap_q_mode)5, (ap_o_mode)3, 0>, 32u>, config2>' to 'dense<array<ap_fixed,50u>,array<ap_fixed<16,6,5,3,0>,32u>,config2>' (firmware/nnet_utils/nnet_dense_stream.h:36:1)
INFO: [HLS 200-111] Finished Architecture Synthesis Time (s): cpu = 00:01:14 ; elapsed = 00:01:18 . Memory (MB): peak = 3065.238 ; gain = 2632.613 ; free physical = 2269 ; free virtual = 7321
INFO: [HLS 200-10] Starting hardware synthesis ...
INFO: [HLS 200-10] Synthesizing 'myproject_axi' ...
WARNING: [SYN 201-103] Legalizing function name 'dense_wrapper<ap_fixed<16, 6, 5, 3, 0>, ap_fixed<16, 6, 5, 3, 0>, config2>' to 'dense_wrapper_ap_fixed_16_6_5_3_0_ap_fixed_16_6_5_3_0_config2_s'.
WARNING: [SYN 201-103] Legalizing function name 'dense<array<ap_fixed,50u>,array<ap_fixed<16,6,5,3,0>,32u>,config2>' to 'dense_array_ap_fixed_50u_array_ap_fixed_16_6_5_3_0_32u_config2_s'.
WARNING: [SYN 201-103] Legalizing function name 'normalize<array<ap_fixed,32u>,array<ap_fixed<16,6,5,3,0>,32u>,config10>' to 'normalize_array_ap_fixed_32u_array_ap_fixed_16_6_5_3_0_32u_config10_s'.
WARNING: [SYN 201-103] Legalizing function name 'relu<array<ap_fixed,32u>,array<ap_ufixed<4,0,4,0,0>,32u>,relu_config4>' to 'relu_array_ap_fixed_32u_array_ap_ufixed_4_0_4_0_0_32u_relu_config4_s'.
WARNING: [SYN 201-103] Legalizing function name 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config5>' to 'dense_wrapper_ap_ufixed_4_0_4_0_0_ap_fixed_16_6_5_3_0_config5_s'.
WARNING: [SYN 201-103] Legalizing function name 'dense<array<ap_ufixed,32u>,array<ap_fixed<16,6,5,3,0>,16u>,config5>' to 'dense_array_ap_ufixed_32u_array_ap_fixed_16_6_5_3_0_16u_config5_s'.
WARNING: [SYN 201-103] Legalizing function name 'normalize<array<ap_fixed,16u>,array<ap_fixed<16,6,5,3,0>,16u>,config11>' to 'normalize_array_ap_fixed_16u_array_ap_fixed_16_6_5_3_0_16u_config11_s'.
WARNING: [SYN 201-103] Legalizing function name 'relu<array<ap_fixed,16u>,array<ap_ufixed<4,0,4,0,0>,16u>,relu_config7>' to 'relu_array_ap_fixed_16u_array_ap_ufixed_4_0_4_0_0_16u_relu_config7_s'.
WARNING: [SYN 201-103] Legalizing function name 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config8>' to 'dense_wrapper_ap_ufixed_4_0_4_0_0_ap_fixed_16_6_5_3_0_config8_s'.
WARNING: [SYN 201-103] Legalizing function name 'dense<array<ap_ufixed,16u>,array<ap_fixed<16,6,5,3,0>,1u>,config8>' to 'dense_array_ap_ufixed_16u_array_ap_fixed_16_6_5_3_0_1u_config8_s'.
WARNING: [SYN 201-103] Legalizing function name 'normalize<array<ap_fixed,1u>,array<ap_fixed<16,6,5,3,0>,1u>,config12>' to 'normalize_array_ap_fixed_1u_array_ap_fixed_16_6_5_3_0_1u_config12_s'.
WARNING: [SYN 201-103] Legalizing function name 'Block_myproject_axi_.exit2622_proc' to 'Block_myproject_axi_exit2622_proc'.
WARNING: [SYN 201-223] Checking resource limit in 'dense_wrapper<ap_fixed<16, 6, 5, 3, 0>, ap_fixed<16, 6, 5, 3, 0>, config2>': cannot find any operation of 'mul'.
WARNING: [SYN 201-223] Checking resource limit in 'normalize<array<ap_fixed,32u>,array<ap_fixed<16,6,5,3,0>,32u>,config10>': cannot find any operation of 'mul'.
WARNING: [SYN 201-223] Checking resource limit in 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config5>': cannot find any operation of 'mul'.
WARNING: [SYN 201-223] Checking resource limit in 'normalize<array<ap_fixed,16u>,array<ap_fixed<16,6,5,3,0>,16u>,config11>': cannot find any operation of 'mul'.
WARNING: [SYN 201-223] Checking resource limit in 'dense_wrapper<ap_ufixed<4, 0, 4, 0, 0>, ap_fixed<16, 6, 5, 3, 0>, config8>': cannot find any operation of 'mul'.
WARNING: [SYN 201-223] Checking resource limit in 'normalize<array<ap_fixed,1u>,array<ap_fixed<16,6,5,3,0>,1u>,config12>': cannot find any operation of 'mul'.

here is my code

# Remove outliers
dataset = dataset[(dataset['Consumption'] >= lower_bound) & (dataset['Consumption'] <= upper_bound)]

# Scale the data
scaler = MinMaxScaler(feature_range=(0, 1))
scaled_data = scaler.fit_transform(dataset[['Consumption']])  # Keep it as a 2D array

# Splitting the data into training, testing, and validation sets
training_size = int(len(scaled_data) * 0.80)
train_data = scaled_data[:training_size]
test_data = scaled_data[training_size:]

val_size = int(len(train_data) * 0.20)
train_data, val_data = train_data[:-val_size], train_data[-val_size:]

# Function to create dataset for Dense model
def create_dataset(dataset, time_step=1):
    dataX, dataY = [], []
    for i in range(len(dataset) - time_step):
        a = dataset[i:(i + time_step), 0]
        dataX.append(a)
        dataY.append(dataset[i + time_step, 0])
    return np.array(dataX), np.array(dataY)

# Creating datasets
time_step = 50
X_train, y_train = create_dataset(train_data, time_step)
X_test, ytest = create_dataset(test_data, time_step)
X_val, yval = create_dataset(val_data, time_step)

# Reshape for Dense input
X_train = X_train.reshape(X_train.shape[0], -1)  # Flatten for Dense input
X_test = X_test.reshape(X_test.shape[0], -1)
X_val = X_val.reshape(X_val.shape[0], -1)

# Create directory for saving files if it doesn't exist
os.makedirs('package', exist_ok=True)

# Save the test data
np.save('package/X_test.npy', X_test)  # Save X_test array
np.save('package/y_test.npy', ytest)   # Save y_test array

# Build the model
model = Sequential([
    # Input layer
    QDense(
        32,  # Reduced neurons
        input_shape=(X_train.shape[1],),
        kernel_quantizer=quantized_bits(4, 0),  # Simplified quantization
        bias_quantizer=quantized_bits(4, 0), 
        kernel_initializer='lecun_uniform',
        kernel_regularizer='l2',
        name='dense_fc1' 
    ),
    QActivation(activation=quantized_relu(4), name='relu1'),
    Dropout(0.2),

    # Hidden layer
    QDense(
        16,  # Further reduced neurons
        kernel_quantizer=quantized_bits(4, 0),  # Simplified quantization
        bias_quantizer=quantized_bits(4, 0),
        kernel_initializer='lecun_uniform',
        name='dense_fc2'  
    ),
    QActivation(activation=quantized_relu(4), name='relu2'),

    # Output layer
    QDense(
        1,
        kernel_quantizer=quantized_bits(4, 0),  # Simplified quantization
        bias_quantizer=quantized_bits(4, 0),
        kernel_initializer='lecun_uniform',
        name='output' 
    )
])


pruning_params = {
    "pruning_schedule": pruning_schedule.ConstantSparsity(0.80, begin_step=2000, frequency=100)
}

# Apply pruning to the model
model = prune.prune_low_magnitude(model, **pruning_params)

# Compile the model
model.compile(optimizer='adam', loss='mean_squared_error')

# Define EarlyStopping callback
early_stopping = EarlyStopping(monitor='val_loss', patience=10, restore_best_weights=True)

# Set up the pruning callback to update pruning steps during training
callbacks = [
    pruning_callbacks.UpdatePruningStep(),
    early_stopping
]

# Fit the model with early stopping and pruning updates
history = model.fit(X_train, y_train, validation_data=(X_val, yval),
                    verbose=1, epochs=10, batch_size=20,
                    callbacks=callbacks)

# Strip the pruning wrappers from the model after training
model = strip_pruning(model)

# Create an HLS config from the Keras model
config = hls4ml.utils.config_from_keras_model(model, granularity='name')

# Set the ReuseFactor for supported layers
for layer in config['LayerName']:
    if 'ReuseFactor' in config['LayerName'][layer]:  # Only update if ReuseFactor exists
        config['LayerName'][layer]['ReuseFactor'] = 64

# Log the configuration for verification
pprint.pprint(config)

# Convert the Keras model to an HLS model
hls_model = hls4ml.converters.convert_from_keras_model(
    model, hls_config=config, io_type='io_stream',
    output_dir='finapynq_hls4ml_prj3', part='xc7z020clg400-1',
    backend='VivadoAccelerator', board='pynq-z2'
)

# Compile the HLS model
hls_model.compile()
hls_model.build(csim=False, export=True, bitfile=True)

Thank you!