Add a converter from dcn_slack_analysis.proto to GViz DataTable format.

PiperOrigin-RevId: 558216192

plugin/tensorboard_plugin_profile/convert/BUILD

@@ -13,6 +13,7 @@ package(
 py_library(
     name = "all_libs",
     deps = [
+        ":dcn_collective_stats_proto_to_gviz",
         ":diagnostics",
         ":input_pipeline_proto_to_gviz",
         ":kernel_stats_proto_to_gviz",
@@ -193,12 +194,35 @@ py_test(
     ],
 )
 
+py_library(
+    name = "dcn_collective_stats_proto_to_gviz",
+    srcs = ["dcn_collective_stats_proto_to_gviz.py"],
+    srcs_version = "PY2AND3",
+    deps = [requirement("gviz_api")],
+)
+
+py_test(
+    name = "dcn_collective_stats_proto_to_gviz_test",
+    size = "small",
+    srcs = ["dcn_collective_stats_proto_to_gviz_test.py"],
+    main = "dcn_collective_stats_proto_to_gviz_test.py",
+    python_version = "PY3",
+    srcs_version = "PY2AND3",
+    deps = [
+        ":dcn_collective_stats_proto_to_gviz",
+        requirement("gviz_api"),
+        "//:expect_tensorflow_installed",
+        "@org_xprof//plugin/tensorboard_plugin_profile/protobuf:protos_all_py_pb2",
+    ],
+)
+
 py_library(
     name = "raw_to_tool_data",
     srcs = ["raw_to_tool_data.py"],
     srcs_version = "PY2AND3",
     visibility = visibility,
     deps = [
+        ":dcn_collective_stats_proto_to_gviz",
         ":input_pipeline_proto_to_gviz",
         ":kernel_stats_proto_to_gviz",
         ":overview_page_proto_to_gviz",

plugin/tensorboard_plugin_profile/convert/dcn_collective_stats_proto_to_gviz.py

@@ -0,0 +1,97 @@
+# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""For conversion of Dcn Collective Stats page protos to GViz DataTables.
+
+Usage:
+    gviz_data_tables = generate_all_chart_tables(dcn_slack_analysis)
+"""
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import sys
+
+import gviz_api
+
+from tensorboard_plugin_profile.protobuf import dcn_slack_analysis_pb2
+
+
+def get_dcn_collective_stats_table_args(dcn_slack_analysis):
+  """Creates a gviz DataTable object from DcnSlackAnalysis proto.
+
+  Args:
+    dcn_slack_analysis: dcn_slack_analysis_pb2.DcnSlackAnalysis.
+
+  Returns:
+    Returns a gviz_api.DataTable
+  """
+
+  table_description = [
+      ("dcnCollectiveName", "string", "Dcn Collective Name"),
+      ("recvOpName", "string", "Recv Op Name"),
+      ("sendOpName", "string", "Send Op Name"),
+      ("slackTime", "number", "Slack Time (ms)"),
+      ("observedDuration", "number", "Observed Duration (ms)"),
+      ("stallDuration", "number", "Stall Duration (ms)"),
+      ("occurrences", "number", "Occurrences"),
+      ("dataTransmittedSize", "number", "Data Transmitted Size"),
+      ("bandwidth", "number", "Bandwidth (Gbps)"),
+  ]
+
+  data = []
+  for slack in dcn_slack_analysis.dcn_slack_summary:
+    row = [
+        slack.rendezvous,
+        slack.recv_op_name,
+        slack.send_op_name,
+        slack.slack_us / 1000,
+        slack.observed_duration_us / 1000,
+        slack.stall_duration_us / 1000,
+        slack.occurrences,
+        slack.bytes_transmitted_over_network,
+    ]
+    if slack.slack_us == 0:
+      row.append(sys.maxsize)
+    else:
+      row.append(
+          (slack.bytes_transmitted_over_network / (slack.slack_us * 8)) / 1000
+      )
+    data.append(row)
+
+  return (table_description, data, [])
+
+
+def generate_dcn_collective_stats_table(dcn_slack_analysis):
+  (table_description, data, custom_properties) = (
+      get_dcn_collective_stats_table_args(dcn_slack_analysis)
+  )
+  return gviz_api.DataTable(table_description, data, custom_properties)
+
+
+def generate_all_chart_tables(dcn_slack_analysis):
+  """Converts a DcnSlackAnalysis proto to gviz DataTables."""
+  return [
+      generate_dcn_collective_stats_table(dcn_slack_analysis),
+  ]
+
+
+def to_json(raw_data):
+  """Converts a serialized DcnCollectiveAnalysis string to json."""
+  dcn_slack_analysis = dcn_slack_analysis_pb2.DcnSlackAnalysis()
+  dcn_slack_analysis.ParseFromString(raw_data)
+  all_chart_tables = generate_all_chart_tables(dcn_slack_analysis)
+  json_join = ",".join(x.ToJSon() if x else "{}" for x in all_chart_tables)
+  return "[" + json_join + "]"

plugin/tensorboard_plugin_profile/convert/dcn_collective_stats_proto_to_gviz_test.py

@@ -0,0 +1,149 @@
+# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+
+"""Tests for dcn_collective_stats_proto_to_gviz."""
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import csv
+import enum
+import io
+
+import gviz_api
+import tensorflow as tf
+
+from tensorboard_plugin_profile.convert import dcn_collective_stats_proto_to_gviz
+from tensorboard_plugin_profile.protobuf import dcn_slack_analysis_pb2
+
+
+class StrEnum(str, enum.Enum):
+  pass
+
+
+class MockValues(StrEnum):
+  DCN_COLLECTIVE_NAME = "collective-1"
+  RECV_OP_NAME = "recv-done"
+  SEND_OP_NAME = "send"
+  SLACK_US = 2
+  OBSERVED_DURATION_US = 12
+  STALL_DURATION_MS = 5
+  OCCURRENCES = 6
+  BYTES_TRANSMITTED_OVER_NETWORK = 8192000
+  BANDWIDTH = 0.512
+
+
+class ProtoToGvizTest(tf.test.TestCase):
+
+  def create_empty_dcn_slack_analysis(self):
+    return dcn_slack_analysis_pb2.DcnSlackAnalysis()
+
+  def create_mock_dcn_slack_summary(self):
+    dcn_slack_summary = dcn_slack_analysis_pb2.DcnSlackSummary(
+        rendezvous=MockValues.DCN_COLLECTIVE_NAME,
+        recv_op_name=MockValues.RECV_OP_NAME,
+        send_op_name=MockValues.SEND_OP_NAME,
+        slack_us=int(MockValues.SLACK_US) * 1000,
+        observed_duration_us=int(MockValues.OBSERVED_DURATION_US) * 1000,
+        stall_duration_us=int(MockValues.STALL_DURATION_MS) * 1000,
+        occurrences=int(MockValues.OCCURRENCES),
+        bytes_transmitted_over_network=int(
+            MockValues.BYTES_TRANSMITTED_OVER_NETWORK
+        )
+    )
+    return dcn_slack_summary
+
+  def create_mock_dcn_slack_analysis(self):
+    dcn_slack_analysis = dcn_slack_analysis_pb2.DcnSlackAnalysis()
+    for _ in range(0, 3):
+      dcn_slack_analysis.dcn_slack_summary.append(
+          self.create_mock_dcn_slack_summary()
+      )
+    return dcn_slack_analysis
+
+  def test_dcn_collective_stats_empty(self):
+    dcn_slack_analysis = self.create_empty_dcn_slack_analysis()
+    data_table = (
+        dcn_collective_stats_proto_to_gviz.generate_dcn_collective_stats_table(
+            dcn_slack_analysis
+        )
+    )
+
+    self.assertEqual(0, data_table.NumberOfRows())
+    self.assertLen(data_table.columns, 9)
+
+  def test_dcn_collective_stats_table(self):
+    dcn_slack_analysis = self.create_mock_dcn_slack_analysis()
+    (table_description, data, custom_properties) = (
+        dcn_collective_stats_proto_to_gviz.get_dcn_collective_stats_table_args(
+            dcn_slack_analysis
+        )
+    )
+    data_table = gviz_api.DataTable(table_description, data, custom_properties)
+
+    self.assertLen(data, 3)
+    self.assertEqual(3, data_table.NumberOfRows())
+    self.assertLen(table_description, 9)
+    self.assertLen(data_table.columns, 9)
+
+    csv_file = io.StringIO(data_table.ToCsv())
+    reader = csv.reader(csv_file)
+
+    expected = [
+        MockValues.DCN_COLLECTIVE_NAME,
+        MockValues.RECV_OP_NAME,
+        MockValues.SEND_OP_NAME,
+        MockValues.SLACK_US,
+        MockValues.OBSERVED_DURATION_US,
+        MockValues.STALL_DURATION_MS,
+        MockValues.OCCURRENCES,
+        MockValues.BYTES_TRANSMITTED_OVER_NETWORK,
+        MockValues.BANDWIDTH,
+    ]
+
+    for rr, row_values in enumerate(reader):
+      if rr == 0:
+        # DataTable columns match schema defined in table_description.
+        for cc, column_header in enumerate(row_values):
+          self.assertEqual(table_description[cc][2], column_header)
+      else:
+        for cc, cell_str in enumerate(row_values):
+          raw_value = data[rr - 1][cc]
+          value_type = table_description[cc][1]
+
+          # Only number and strings are used in the DataTable schema.
+          self.assertIn(value_type, ["number", "string"])
+
+          # Encode in similar fashion as DataTable.ToCsv().
+          expected_value = gviz_api.DataTable.CoerceValue(raw_value, value_type)
+          self.assertNotIsInstance(expected_value, tuple)
+          self.assertEqual(expected_value, raw_value)
+
+          # Check against expected values we have set in our mock table.
+          if value_type == "string":
+            self.assertEqual(expected[cc], cell_str)
+          else:
+            if (
+                expected[cc] == MockValues.OCCURRENCES
+                or expected[cc] == MockValues.BYTES_TRANSMITTED_OVER_NETWORK
+            ):
+              self.assertEqual(str(int(expected[cc])), cell_str)
+            else:
+              self.assertEqual(str(float(expected[cc])), cell_str)
+
+
+if __name__ == "__main__":
+  tf.test.main()

plugin/tensorboard_plugin_profile/convert/raw_to_tool_data.py

@@ -26,6 +26,7 @@
 import logging
 
 from tensorflow.python.profiler.internal import _pywrap_profiler  # pylint: disable=g-direct-tensorflow-import
+from tensorboard_plugin_profile.convert import dcn_collective_stats_proto_to_gviz
 from tensorboard_plugin_profile.convert import input_pipeline_proto_to_gviz
 from tensorboard_plugin_profile.convert import kernel_stats_proto_to_gviz
 from tensorboard_plugin_profile.convert import overview_page_proto_to_gviz
@@ -167,6 +168,11 @@ def xspace_to_tool_data(
     raw_data, success = xspace_wrapper_func(xspace_paths, tool, options)
     if success:
       data = raw_data
+  elif tool == 'dcn_collective_stats':
+    options = {'host_name': params.get('host')}
+    raw_data, success = xspace_wrapper_func(xspace_paths, tool, options)
+    if success:
+      data = dcn_collective_stats_proto_to_gviz.to_json(raw_data)
   else:
     logger.warning('%s is not a known xplane tool', tool)
   return data, content_type

plugin/tensorboard_plugin_profile/profile_plugin.py

@@ -114,6 +114,7 @@
     'overview_page^',
     'pod_viewer^',
     'tf_data_bottleneck_analysis^',
+    'dcn_collective_stats^',
 ])
 
 # XPlane generated tools that only support all host mode.

plugin/tensorboard_plugin_profile/protobuf/BUILD

@@ -10,6 +10,7 @@ package(
 proto_library(
     name = "protos_all",
     srcs = [
+        "dcn_slack_analysis.proto",
         "diagnostics.proto",
         "input_pipeline.proto",
         "kernel_stats.proto",
@@ -24,6 +25,7 @@ proto_library(
 py_proto_library(
     name = "protos_all_py_pb2",
     srcs = [
+        "dcn_slack_analysis.proto",
         "diagnostics.proto",
         "input_pipeline.proto",
         "kernel_stats.proto",