Adds one example to benchmark onnx serialization functions

_doc/examples/plot_benchmark_onnx_serialize.py

@@ -0,0 +1,110 @@
+"""
+.. _l-benchmark-onnx-serialize:
+
+SerializeToString and ParseFromString
+=====================================
+
+
+.. contents::
+    :local:
+
+Startup
++++++++
+
+This section creates an ONNX graph if there is not one.
+
+"""
+import numpy
+import onnx
+from cpyquickhelper.numbers.speed_measure import measure_time
+import matplotlib.pyplot as plt
+import pandas
+from tqdm import tqdm
+from mlprodict.testing.experimental_c_impl.experimental_c import code_optimisation
+from mlprodict.plotting.text_plot import onnx_simple_text_plot
+from mlprodict.npy.xop import loadop
+
+
+############################################
+# Available optimisation on this machine.
+
+print(code_optimisation())
+
+
+##############################
+# Build an ONNX graph of different size
+# +++++++++++++++++++++++++++++++++++++
+
+def build_model(n_nodes, size, opv=15):
+    OnnxAdd, OnnxIdentity = loadop('Add', 'Identity')
+    x = 'X'
+    for n in range(n_nodes):
+        y = OnnxAdd(x, numpy.random.randn(size).astype(numpy.float32),
+                    op_version=opv)
+        x = y
+    final = OnnxIdentity(x, op_version=opv, output_names=['Y'])
+    return final.to_onnx(numpy.float32, numpy.float32, target_opset=opv)
+
+
+model = build_model(2, 5)
+print(onnx_simple_text_plot(model))
+
+##########################################
+# Measure the time of serialization functions
+# +++++++++++++++++++++++++++++++++++++++++++
+
+
+def parse(buffer):
+    proto = onnx.ModelProto()
+    proto.ParseFromString(buffer)
+    return proto
+
+
+data = []
+nodes = [5, 10, 20]
+for size in tqdm([10, 100, 1000, 10000, 100000, 200000, 300000]):
+    for n_nodes in nodes:
+        onx = build_model(n_nodes, size)
+        serialized = onx.SerializeToString()
+        onnx_size = len(serialized)
+        obs = measure_time(lambda: onx.SerializeToString(),
+                           div_by_number=True, repeat=20)
+        obs['size'] = size
+        obs['n_nodes'] = n_nodes
+        obs['onnx_size'] = onnx_size
+        obs['task'] = "SerializeToString"
+        data.append(obs)
+
+        parsed = parse(serialized)
+        obs = measure_time(lambda: parse(serialized),
+                           div_by_number=True, repeat=20)
+        obs['size'] = size
+        obs['n_nodes'] = n_nodes
+        obs['onnx_size'] = onnx_size
+        obs['task'] = "ParseFromString"
+        data.append(obs)
+
+
+df = pandas.DataFrame(data).sort_values(
+    ['task', 'onnx_size', 'size', 'n_nodes'])
+df[['task', 'onnx_size', 'size', 'n_nodes', 'average']]
+
+
+###############################################
+# Summary
+# +++++++
+
+piv = df.pivot(index='onnx_size', columns='task', values='average')
+piv
+
+#########################################
+# Graph
+# +++++
+
+fig, ax = plt.subplots(1, 1)
+piv.plot(title="Time processing of serialization functions\n"
+               "lower better", ax=ax)
+ax.set_xlabel("onnx size")
+ax.set_ylabel("s")
+
+# plt.show()

_doc/sphinxdoc/source/api/onnx_python/serialize.rst

@@ -65,3 +65,10 @@ an array of bytes.
 
 .. autoclass:: onnx.TensorProto
     :members: SerializeToString
+
+Performance
++++++++++++
+
+.. toctree::
+
+    ../../gyexamples/plot_benchmark_onnx_serialize

_doc/sphinxdoc/source/api/onnx_python/utils.rst

@@ -44,4 +44,4 @@ dimension
 
 .. autofunction:: onnx.compose.expand_out_dim_graph
 
-.. autofunction:: onnx.tools.update_model_dims
+.. autofunction:: onnx.tools.update_model_dims

_doc/sphinxdoc/source/api/onnxruntime_python/inference.rst

@@ -83,4 +83,3 @@ GraphOptimizationLevel
 
 .. autoclass:: onnxruntime.capi._pybind_state.GraphOptimizationLevel
     :members:
-