Ctrl+K

Output Monitoring

Output Monitoring#

How to Read The Output from Hidden Layers.#

Basically, we can read all 9 layers output. The CnnLayerConfig of the Samna Configuration object has a boolean attribute called monitor_enable. If you want to read a specific layer’s output, you must set the monitor_enable as True for that layer.

By setting multiple layers’ monitor_enable as True, we will receive samna.xxx.event.Spike events from the output buffer. Each samna.xxx.event.Spike has a layer attribute to indicate which layer it comes from.

from torch import nn
from sinabs.layers import IAFSqueeze
from sinabs.backend.dynapcnn import DynapcnnNetwork
import samna

SNN = nn.Sequential(

    nn.Conv2d(1, 2, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),

    nn.Conv2d(2, 2, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),

    nn.Conv2d(2, 4, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),
)

"""
1. you can manually set the hidden layers monitor_enable as True
""" 
dynapcnn = DynapcnnNetwork(snn=SNN, input_shape=(1, 16, 16), dvs_input=False)
samna_cfg = dynapcnn.make_config(device="speck2fmodule")

hidden_layers = [0, 1]
for h_layer in hidden_layers:
    samna_cfg.cnn_layers[h_layer].monitor_enable = True

# check if the all layer's output can be read
# by default, we will monitor last layer's output
for layer in [0, 1, 2]:
    print(samna_cfg.cnn_layers[layer].monitor_enable)

# finally we just need to apply the samna configuration to the devkit, we finish the deployment.
devkit = samna.device.open_device("Speck2fModuleDevKit")
devkit.get_model().apply_configuration(samna_cfg)


"""
2. We can use a more convenient API (recommended)
"""
dynapcnn = DynapcnnNetwork(snn=SNN, input_shape=(1, 16, 16), dvs_input=False)
dynapcnn.to(device="speck2fmodule", monitor_layers=[0, 1, 2])

How to Read The Vmem of the Neurons#

Samna provide a type of event called samna.xxx.event.ReadNeuronValue to help us read the membrane potential of neurons. Here we give an example on Speck2f.

from torch import nn
from sinabs.layers import IAFSqueeze
from sinabs.backend.dynapcnn import DynapcnnNetwork
from sinabs.backend.dynapcnn.io import calculate_neuron_address, neuron_address_to_cxy
import samna

SNN = nn.Sequential(

    nn.Conv2d(1, 2, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),

    nn.Conv2d(2, 2, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),

    nn.Conv2d(2, 4, kernel_size=(1, 1), bias=False),
    IAFSqueeze(batch_size=1, min_v_mem=-1.0),
)

# deploy the snn to the devkit
dynapcnn = DynapcnnNetwork(snn=SNN, input_shape=(1, 16, 16), dvs_input=False)
dynapcnn.to(device="speck2fmodule")

# create ReadNeuronValue events as input
input_events = []

layers = [0, 1, 2]
x_list = [1, 1, 2]
y_list = [1, 1, 2]
channel = 0
for layer, x, y in zip(layers, x_list, y_list):
    address = calculate_neuron_address(x=x, y=y, c=channel, feature_map_size=(2, 16, 16))
    ev = samna.speck2f.event.ReadNeuronValue()
    ev.address = address
    ev.layer = layer
    input_events.append(ev)

# send the input to the devkit
output_events = dynapcnn(input_events)

# check the output 
for ev in output_events:

    address = ev.address
    c, x, y = neuron_address_to_cxy(address, feature_map_size=(2, 16, 16))
    layer = ev.layer
    vmem = ev.neuron_state
    print(f"vmem of layer {layer} at channel={c}, x={x}, y={y} is {vmem}!")
Contents