improve the full bnn testbench

Use numpy functions, because they are easier to use.

sim/test_bnn.py

@@ -24,21 +24,18 @@
 async def run_test(dut):
  height = dut.C_INPUT_HEIGHT.value.integer
  width = dut.C_INPUT_WIDTH.value.integer
- input_image = [i % 255 for i in range(height * width)]
+ input_image = np.random.randint(0, 255, (1, height, width, 1), dtype=np.uint8)
 
  model = tf.keras.models.load_model("../../models/test")
  lq.models.summary(model)
- data = np.reshape(np.array(input_image), (1, height, width, 1))
 
  # https://keras.io/getting_started/faq/#how-can-i-obtain-the-output-of-an-intermediate-layer-feature-extraction
  extractor = tf.keras.Model(
  inputs=model.inputs, outputs=[layer.output for layer in model.layers]
  )
- features = extractor(data)
- class_scores_pos = list(features[-2].numpy().flat)
- class_scores = [
- round((score + features[-4].shape[-1]) / 2) for score in class_scores_pos
- ]
+ features = extractor(input_image)
+ # class scores = (output before softmax + fan in) / 2
+ class_scores_pos = np.around((features[-2].numpy() + features[-4].shape[-1]) / 2)
 
  output_bitwitdh = dut.C_OUTPUT_CHANNEL_BITWIDTH.value.integer
  output_mon = ImageMonitor(
@@ -57,21 +54,21 @@ async def run_test(dut):
  dut.isl_valid <= 0
  await tick.wait()
 
- for pixel in input_image:
+ for pixel in input_image.flat:
  dut.isl_valid <= 1
- dut.islv_data <= pixel
+ dut.islv_data <= int(pixel)
  await tick.wait()
  dut.isl_valid <= 0
  await tick.wait()
  await tick.wait()
 
  await tick.wait_multiple(height * width)
 
- print(class_scores_pos)
- print(class_scores)
- print(output_mon.output)
- print(features)
- assert output_mon.output == class_scores
+ np.testing.assert_almost_equal(
+  np.resize(np.array(output_mon.output), class_scores_pos.shape),
+  class_scores_pos,
+  decimal=0,
+ )
 
 
 def test_bnn():