Revert TRT4 support

This reverts commit dc693c74b179800f19d376ff79857f95e3b637b0.

tensorflow/compiler/tf2tensorrt/convert/convert_nodes.cc

@@ -1176,7 +1176,6 @@ Status TrtNodeValidator::ConvertConstToWeights(
   return status;
 }
 
-#if IS_TRT_VERSION_GE(5, 0, 0, 0)
 static void InitializeTrtPlugins() {
   static mutex plugin_mutex(LINKER_INITIALIZED);
   static bool plugin_initialized = false;
@@ -1209,16 +1208,13 @@ static void InitializeTrtPlugins() {
     }
   }
 }
-#endif
 
 Converter::Converter(nvinfer1::INetworkDefinition* trt_network,
                      TrtPrecisionMode precision_mode, bool use_calibration)
     : trt_network_(trt_network),
       precision_mode_(precision_mode),
       use_calibration_(use_calibration) {
-#if IS_TRT_VERSION_GE(5, 0, 0, 0)
   InitializeTrtPlugins();
-#endif
   this->RegisterOpConverters();
 }
 
@@ -4700,9 +4696,6 @@ Status ConvertMatMulHelper(OpConverterParams* params,
       !transpose_a && input_a.is_tensor() && input_b.is_weights();
   const bool should_use_fc = can_use_fc && input_a.GetTrtDims().nbDims >= 3 &&
                              input_b.GetTrtDims().nbDims == 2;
-  // If TRT < 5, a fully connected layer must be used as the matmul op is
-  // unsupported
-#if IS_TRT_VERSION_GE(5, 0, 0, 0)
   // If int8 is specified, FC must be used unless it is not compatible, as MM
   // does not support int8 at this time.
   if (should_use_fc || (can_use_fc && params->converter->precision_mode() ==
@@ -4753,10 +4746,6 @@ Status ConvertMatMulHelper(OpConverterParams* params,
   nvinfer1::ITensor* output_tensor = layer->getOutput(0);
   params->outputs->push_back(TRT_TensorOrWeights(output_tensor));
   return Status::OK();
-#else
-  return ConvertFullyConnectedHelper(
-      params, input_a.tensor(), input_b.weights(), transpose_b, node_name);
-#endif
 }
 
 // inputs are both two dimensional (ops::MatMul)

tensorflow/python/compiler/tensorrt/test/tf_trt_integration_test_base.py

@@ -30,7 +30,6 @@
 import numpy as np
 import six
 
-from tensorflow.compiler.tf2tensorrt.wrap_py_utils import get_linked_tensorrt_version
 from tensorflow.compiler.tf2tensorrt.wrap_py_utils import is_tensorrt_enabled
 from tensorflow.core.framework import graph_pb2
 from tensorflow.core.protobuf import config_pb2
@@ -56,7 +55,6 @@
 from tensorflow.python.training.tracking import tracking
 from tensorflow.python.util import nest
 
-
 TfTrtIntegrationTestParams = namedtuple(
     "TfTrtIntegrationTestParams",
     [
@@ -91,11 +89,7 @@
 FP32 = "FP32"
 FP16 = "FP16"
 INT8 = "INT8"
-TRT_VERSION = get_linked_tensorrt_version()
-if TRT_VERSION >= (5, 0, 0):
-  PRECISION_MODES = [FP32, FP16, INT8]
-else:
-  PRECISION_MODES = [FP32, FP16]
+PRECISION_MODES = [FP32, FP16, INT8]
 
 
 def IsQuantizationMode(mode):
@@ -143,16 +137,13 @@ def OptimizerDisabledRewriterConfig():
 class TfTrtIntegrationTestBase(test_util.TensorFlowTestCase):
   """Class to test Tensorflow-TensorRT integration."""
 
-  TRT_VERSION = get_linked_tensorrt_version()
   @property
   def trt_incompatible_op(self):
     return math_ops.erf
 
   @property
   def precision_modes(self):
-    if TRT_VERSION >= (5, 0, 0):
-      return ["FP32", "FP16", "INT8"]
-    return ["FP32", "FP16"]
+    return ["FP32", "FP16", "INT8"]
 
   # str is bytes in py2, but unicode in py3.
   def _ToUnicode(self, s):
@@ -791,17 +782,12 @@ def _GetTestConfigsV1():
   # whether to run specific ones with ShouldRunTest().
   #
   # Note: INT8 without calibration behaves like FP32/FP16.
-  if TRT_VERSION >= (5, 0, 0):
-    opts = list(
+  opts = list(
       itertools.product([FP32, FP16, INT8], [convert_online, convert_offline],
                         [dynamic_engine, static_engine], [no_calibration]))
-    # We always run calibration with offline tool.
-    # TODO(aaroey): static calibration engine is not supported yet.
-    opts.append((INT8, convert_offline, dynamic_engine, use_calibration))
-  else:
-    opts = list(
-      itertools.product([FP32, FP16], [convert_online, convert_offline],
-                        [dynamic_engine, static_engine], [no_calibration]))
+  # We always run calibration with offline tool.
+  # TODO(aaroey): static calibration engine is not supported yet.
+  opts.append((INT8, convert_offline, dynamic_engine, use_calibration))
   return opts
 
 
@@ -822,17 +808,12 @@ def _GetTestConfigsV2():
   # - For simplicity we don't test online conversion which requires setting the
   #   Grappler config in default eager context.
   # - INT8 without calibration behaves like FP32/FP16.
-  if TRT_VERSION >= (5, 0, 0):
-    opts = list(
+  opts = list(
       itertools.product([FP32, FP16, INT8], [convert_offline], [dynamic_engine],
                         [no_calibration]))
-    # We always run calibration with offline tool.
-    # TODO(aaroey): INT8+calibration is not supported yet in V2.
-    # opts.append((INT8, convert_offline, dynamic_engine, use_calibration))
-  else:
-    opts = list(
-      itertools.product([FP32, FP16], [convert_offline], [dynamic_engine],
-                        [no_calibration]))
+  # We always run calibration with offline tool.
+  # TODO(aaroey): INT8+calibration is not supported yet in V2.
+  # opts.append((INT8, convert_offline, dynamic_engine, use_calibration))
   return opts
 
 

tensorflow/python/compiler/tensorrt/trt_convert_test.py

@@ -24,7 +24,6 @@
 
 import numpy as np
 
-from tensorflow.compiler.tf2tensorrt.wrap_py_utils import get_linked_tensorrt_version
 from tensorflow.compiler.tf2tensorrt.wrap_py_utils import is_tensorrt_enabled
 from tensorflow.core.framework import graph_pb2
 from tensorflow.core.protobuf import config_pb2
@@ -77,22 +76,13 @@ def testGetTensorrtRewriterConfig(self):
     """Test case for TrtGraphConverter.get_tensorrt_rewriter_config()."""
     if not is_tensorrt_enabled():
       return
-    TRT_VERSION = get_linked_tensorrt_version()
-    if TRT_VERSION >= (5, 0, 0):
-      conversion_params = trt_convert.DEFAULT_TRT_CONVERSION_PARAMS._replace(
+    conversion_params = trt_convert.DEFAULT_TRT_CONVERSION_PARAMS._replace(
         max_batch_size=128,
         max_workspace_size_bytes=1234,
         precision_mode="INT8",
         minimum_segment_size=10,
         is_dynamic_op=True,
         maximum_cached_engines=2)
-    else:
-      conversion_params = trt_convert.DEFAULT_TRT_CONVERSION_PARAMS._replace(
-        max_batch_size=128,
-        max_workspace_size_bytes=1234,
-        minimum_segment_size=10,
-        is_dynamic_op=True,
-        maximum_cached_engines=2)
     rewriter_cfg = trt_convert.get_tensorrt_rewriter_config(
         conversion_params=conversion_params)
     self.assertEqual(["constfold", "layout", "constfold"],
@@ -115,8 +105,7 @@ def testGetTensorrtRewriterConfig(self):
     self.assertEqual(True, trt_optimizer.parameter_map["is_dynamic_op"].b)
     self.assertEqual(1234,
                      trt_optimizer.parameter_map["max_workspace_size_bytes"].i)
-    if TRT_VERSION >= (5, 0, 0):
-      self.assertEqual(
+    self.assertEqual(
         trt_convert._to_bytes("INT8"),
         trt_optimizer.parameter_map["precision_mode"].s)
     self.assertEqual(2, trt_optimizer.parameter_map["maximum_cached_engines"].i)
@@ -311,13 +300,7 @@ def testTrtGraphConverter_BasicConversion(self):
     input_saved_model_dir = self.mkdtemp()
     self._WriteInputSavedModel(input_saved_model_dir)
 
-    TRT_VERSION = get_linked_tensorrt_version()
-    if TRT_VERSION >= (5, 0, 0):
-      calibration_opts = [False, True]
-    else:
-      calibration_opts = [False]
-
-    for need_calibration in calibration_opts:
+    for need_calibration in [False, True]:
       # Use GraphDef as input.
       self._TestTrtGraphConverter()