Quartus Extensions

hls4ml/backends/quartus/quartus_backend.py

@@ -43,29 +43,32 @@ def _register_flows(self):
         ]
         quantization_flow = register_flow('quantization', quantization_passes, requires=[init_flow], backend=self.name)
 
+        optimization_passes = []
+        optimization_flow = register_flow('optimize', optimization_passes, requires=[init_flow], backend=self.name)
+
         templates = self._get_layer_templates()
-        template_flow = register_flow('apply_templates', templates, requires=[init_flow], backend=self.name)
+        template_flow = register_flow('apply_templates', self._get_layer_templates, requires=[init_flow], backend=self.name)
 
         writer_passes = [
             'make_stamp',
             'quartus:write_hls'
         ]
-        writer_flow_requirements = ['optimize', quartus_types_flow, template_flow]
-        self._writer_flow = register_flow('write', writer_passes, requires=writer_flow_requirements, backend=self.name)
+
+        self._writer_flow = register_flow('write', writer_passes, requires=['quartus:ip'], backend=self.name)
 
         all_passes = get_backend_passes(self.name)
 
         extras = [
             # Ideally this should be empty
-            opt_pass for opt_pass in all_passes if opt_pass not in initializers + quartus_types + templates + writer_passes
+            opt_pass for opt_pass in all_passes if opt_pass not in initializers + streaming_passes + quartus_types + quantization_passes + templates + optimization_passes + writer_passes
         ]
 
         if len(extras) > 0:
             extras_flow = register_flow('extras', extras, requires=[init_flow], backend=self.name)
         else:
             extras_flow = None
 
-        ip_flow_requirements = ['optimize', init_flow, streaming_flow, quantization_flow, quartus_types_flow, extras_flow, template_flow]
+        ip_flow_requirements = ['optimize', init_flow, streaming_flow, quantization_flow, optimization_flow, quartus_types_flow, extras_flow, template_flow]
         ip_flow_requirements = list(filter(None, ip_flow_requirements))
 
         self._default_flow = register_flow('ip', None, requires=ip_flow_requirements, backend=self.name)

test/pytest/test_extensions.py

@@ -1,13 +1,12 @@
+import pytest
 import hls4ml
-import tensorflow as tf
 import numpy as np
-import pytest
+import tensorflow as tf
 from pathlib import Path
 
 test_root_path = Path(__file__).parent
 
 # Keras implementation of a custom layer
-
 class KReverse(tf.keras.layers.Layer):
     ''' Keras implementation of a hypothetical custom layer '''
     def __init__(self):
@@ -16,8 +15,7 @@ def __init__(self):
     def call(self, inputs):
         return tf.reverse(inputs, axis=[-1])
 
-# hls4ml implementations
-
+# hls4ml layer implementation
 class HReverse(hls4ml.model.layers.Layer):
     ''' hls4ml implementation of a hypothetical custom layer '''
 
@@ -27,8 +25,35 @@ def initialize(self):
         dims = inp.dim_names
         self.add_output_variable(shape, dims)
 
+# hls4ml optimizer to remove duplicate optimizer
+class RemoveDuplicateReverse(hls4ml.model.optimizer.OptimizerPass):
+    '''OptimizerPass to remove consecutive HReverse layers.'''
+
+    def match(self, node):
+        return isinstance(node, HReverse) and \
+               isinstance(node.get_input_node(), HReverse)
+
+    def transform(self, model, node):
+        first = node.get_input_node()
+        second = node
 
-# Templates
+        model.remove_node(first, rewire=True)
+        model.remove_node(second, rewire=True)
+        return True
+
+# Parser for converter
+def parse_reverse_layer(keras_layer, input_names, input_shapes, data_reader, config):
+    layer = {}
+    layer['class_name'] = 'HReverse'
+    layer['name'] = keras_layer['config']['name']
+    layer['n_in'] = input_shapes[0][1]
+
+    if input_names is not None:
+        layer['inputs'] = input_names
+
+    return layer, [shape for shape in input_shapes[0]]
+
+# HLS Templates - No specific pragmas used; generic enough for both Intel and Vivado
 
 rev_config_template = """struct config{index} : nnet::reverse_config {{
     static const unsigned n_in = {n_in};
@@ -55,8 +80,6 @@ def format(self, node):
         params = self._default_function_params(node)
         return self.template.format(**params)
 
-
-# HLS implementation
 rev_hls = \
 """#ifndef NNET_REVERSE_H_
 #define NNET_REVERSE_H_
@@ -74,8 +97,6 @@ def format(self, node):
     data_T input[CONFIG_T::n_in],
     data_T reversed[CONFIG_T::n_in]
 ) {
-    #pragma HLS PIPELINE
-
     for (int i = 0; i < CONFIG_T::n_in; i++) {
         reversed[CONFIG_T::n_in - 1 - i] = input[i];
     }
@@ -86,43 +107,19 @@ def format(self, node):
 #endif
 """
 
-class RemoveDuplicateReverse(hls4ml.model.optimizer.OptimizerPass):
-    '''OptimizerPass to remove consecutive HReverse layers.'''
-
-    def match(self, node):
-        return isinstance(node, HReverse) and \
-               isinstance(node.get_input_node(), HReverse)
-
-    def transform(self, model, node):
-        first = node.get_input_node()
-        second = node
-
-        model.remove_node(first, rewire=True)
-        model.remove_node(second, rewire=True)
-        return True
-
-# Parser for converter
-def parse_reverse_layer(keras_layer, input_names, input_shapes, data_reader, config):
-    layer = {}
-    layer['class_name'] = 'HReverse'
-    layer['name'] = keras_layer['config']['name']
-    layer['n_in'] = input_shapes[0][1]
-
-    if input_names is not None:
-        layer['inputs'] = input_names
-
-    return layer, [shape for shape in input_shapes[0]]
-
-def test_extensions(tmp_path):
+@pytest.fixture(scope='session', autouse=True)
+def regsister_custom_layer():
     # Register the converter for custom Keras layer
     hls4ml.converters.register_keras_layer_handler('KReverse', parse_reverse_layer)
 
     # Register the hls4ml's IR layer
     hls4ml.model.layers.register_layer('HReverse', HReverse)
 
+@pytest.mark.parametrize('backend_id', ['Vivado', 'Quartus'])
+def test_extensions(tmp_path, backend_id):
     # Register the optimization passes (if any)
-    backend = hls4ml.backends.get_backend('Vivado')
-    backend.register_pass('remove_duplicate_reverse', RemoveDuplicateReverse, flow='vivado:optimize')
+    backend = hls4ml.backends.get_backend(backend_id)
+    backend.register_pass('remove_duplicate_reverse', RemoveDuplicateReverse, flow=f'{backend_id.lower()}:optimize')
 
     # Register template passes for the given backend
     backend.register_template(HReverseConfigTemplate)
@@ -148,15 +145,15 @@ def test_extensions(tmp_path):
 
     hmodel = hls4ml.converters.convert_from_keras_model(
         kmodel,
-        output_dir=str(test_root_path / 'hls4mlprj_extensions'),
-        backend='Vivado',
+        output_dir=str(test_root_path / f'hls4mlprj_extensions_{backend_id}'),
+        backend=backend_id,
         io_type='io_parallel',
-        hls_config={ 'Model': { 'Precision': 'ap_int<4>', 'ReuseFactor': 1} })
+        hls_config={ 'Model': { 'Precision': 'ap_int<6>', 'ReuseFactor': 1} })
 
     hmodel.compile()
     hres = hmodel.predict(x.astype('float32'))
 
     # Check if the optimizer pass was applied
-    assert 'vivado:remove_duplicate_reverse' in hmodel._applied_flows[0]['vivado:optimize']
+    assert f'{backend_id.lower()}:remove_duplicate_reverse' in hmodel._applied_flows[0][f'{backend_id.lower()}:optimize']
 
     np.testing.assert_array_equal(kres, hres)