remove fcplugin to accelerate pipeline run time

Signed-off-by: Rajeev Rao <rajeevrao@nvidia.com>

demo/BERT/builder.py

@@ -54,7 +54,6 @@
 emln_plg_creator = plg_registry.get_plugin_creator("CustomEmbLayerNormPluginDynamic", "1", "")
 qkv2_plg_creator = plg_registry.get_plugin_creator("CustomQKVToContextPluginDynamic", "1", "")
 skln_plg_creator = plg_registry.get_plugin_creator("CustomSkipLayerNormPluginDynamic", "1", "")
-fc_plg_creator = plg_registry.get_plugin_creator("CustomFCPluginDynamic", "1", "")
 
 """
 Attentions Keys
@@ -214,16 +213,6 @@ def skipln(prefix, config, init_dict, network, input_tensor, skip, bias=None):
     layer = network.add_plugin_v2(skipln_inputs, skipln_plug)
     return layer
 
-def custom_fc(config, network, input_tensor, out_dims, W):
-    pf_out_dims = trt.PluginField("out_dims", np.array([out_dims], dtype=np.int32), trt.PluginFieldType.INT32)
-    pf_W = trt.PluginField("W", W.numpy(), trt.PluginFieldType.FLOAT32)
-    pf_type = trt.PluginField("type_id", np.array([1 if config.use_fp16 else 0], np.int32), trt.PluginFieldType.INT32)
-    pfc = trt.PluginFieldCollection([pf_out_dims, pf_W, pf_type])
-    fc_plugin = fc_plg_creator.create_plugin("fcplugin", pfc)
-    plug_inputs = [input_tensor]
-    out_dense = network.add_plugin_v2(plug_inputs, fc_plugin)
-    return out_dense
-
 def transformer_layer_opt(prefix, config, init_dict, network, input_tensor, imask):
     """
     Add the transformer layer
@@ -243,10 +232,9 @@ def transformer_layer_opt(prefix, config, init_dict, network, input_tensor, imas
 
     # FC0
     B_aout = init_dict[prefix + B_AOUT]
+    W_aout = init_dict[prefix + W_AOUT]
     if config.use_int8:
-        W_aout = init_dict[prefix + W_AOUT]
         attention_out_fc = network.add_convolution(attention_heads, hidden_size, (1, 1), W_aout, B_aout)
-        B_aout = None
 
         if not config.use_int8_skipln:
             attention_out_fc.set_output_type(0, trt.DataType.HALF if config.use_fp16 else trt.DataType.FLOAT)
@@ -255,10 +243,9 @@ def transformer_layer_opt(prefix, config, init_dict, network, input_tensor, imas
             dr_fc_aout = init_dict[prefix + 'attention_output_add_local_input_quantizer_amax']
             set_output_range(attention_out_fc, dr_fc_aout)
     else:
-        W_aoutT = init_dict[prefix + W_AOUT + "_notrans"]
-        attention_out_fc = custom_fc(config, network, attention_heads, hidden_size, W_aoutT)
+        attention_out_fc = network.add_fully_connected(attention_heads, hidden_size, W_aout, B_aout)
 
-    skiplayer = skipln(prefix + "attention_output_layernorm_",config, init_dict, network, attention_out_fc.get_output(0), input_tensor, B_aout)
+    skiplayer = skipln(prefix + "attention_output_layernorm_",config, init_dict, network, attention_out_fc.get_output(0), input_tensor, None)
     attention_ln = skiplayer.get_output(0)
     if config.use_qat:
         dr_skln1 = init_dict[prefix + 'intermediate_dense_input_amax']
@@ -303,23 +290,21 @@ def transformer_layer_opt(prefix, config, init_dict, network, input_tensor, imas
     # FC2
     # Dense to hidden size
     B_lout = init_dict[prefix + B_LOUT]
+    W_lout = init_dict[prefix + W_LOUT]
     if config.use_int8 and not config.use_fc2_gemm:
-        W_lout = init_dict[prefix + W_LOUT]
         out_dense = network.add_convolution(intermediate_act, hidden_size, (1, 1), W_lout, B_lout)
-        B_lout = None
 
         if not config.use_int8_skipln:
             out_dense.set_output_type(0, trt.DataType.HALF if config.use_fp16 else trt.DataType.FLOAT)
     else:
-        W_loutT = init_dict[prefix + W_LOUT + "_notrans"]
-        out_dense = custom_fc(config, network, intermediate_act, hidden_size, W_loutT)
+        out_dense = network.add_fully_connected(intermediate_act, hidden_size, W_lout, B_lout)
 
     if config.use_qat:
         dr_fc_out = init_dict[prefix + 'output_add_local_input_quantizer_amax']
         set_output_range(out_dense, dr_fc_out)
     set_output_name(out_dense, prefix + "output_", "dense")
 
-    out_layer = skipln(prefix + "output_layernorm_", config, init_dict, network, out_dense.get_output(0), attention_ln, B_lout)
+    out_layer = skipln(prefix + "output_layernorm_", config, init_dict, network, out_dense.get_output(0), attention_ln, None)
     set_output_name(out_layer, prefix + "output_", "reshape")
 
     return out_layer