Find back net_config_params for VGG.

codeGenerator/examples/VGG16/net_config_params.txt

@@ -1,16 +1,18 @@
-Network Structure: Input Convolution ReLU MaxPooling Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat MaxPooling Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat MaxPooling Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Convolution ReLU Convolution ReLU Convolution(padding) ReLU Concat Dropout Convolution ReLU GlobalAvePooling Softmax
-nn_in_data_size_conv: 227 56 56 56 56 56 56 28 28 28 28 28 28 14 14 14 14 14 14 14 14 14 14 14 14 14
-nn_channel_size_conv: 3 1 1 3 1 1 3 1 1 3 1 1 3 1 1 3 1 1 3 1 1 3 1 1 3 1
-nn_padding_conv: 0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 1 0 0 1 0 0 1 0 0 1 0
-nn_stride_conv: 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
-nn_in_number_conv: 3 64 16 16 128 16 16 128 32 32 256 32 32 256 48 48 384 48 48 384 64 64 512 64 64 512
-nn_out_number_conv: 64 16 64 64 16 64 64 32 128 128 32 128 128 48 192 192 48 192 192 64 256 256 64 256 256 1000
-nn_group_conv: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
-nn_bias_conv: 64 16 64 64 16 64 64 32 128 128 32 128 128 48 192 192 48 192 192 64 256 256 64 256 256 1000
-nn_in_data_size_pooling: 113 56 28 14
-nn_channel_size_pooling: 3 3 3 0
-nn_padding_pooling: 0 0 0 0
-nn_stride_pooling: 2 2 2 1
-nn_in_number_pooling: 64 128 256 1000
-nn_in_number_concat: 64 64 128 128 192 192 256 256
-nn_input_size_concat: 56 56 28 28 14 14 14 14
+Network Structure: Convolution(padding) ReLU Convolution(padding) ReLU MaxPooling Convolution(padding) ReLU Convolution(padding) ReLU MaxPooling Convolution(padding) ReLU Convolution(padding) ReLU Convolution(padding) ReLU MaxPooling Convolution(padding) ReLU Convolution(padding) ReLU Convolution(padding) ReLU MaxPooling Convolution(padding) ReLU Convolution(padding) ReLU Convolution(padding) ReLU MaxPooling InnerProduct ReLU Dropout InnerProduct ReLU Dropout InnerProduct Softmax
+ nn_in_data_size_conv: 224 224 112 112 56 56 56 28 28 28 14 14 14
+ nn_channel_size_conv: 3 3 3 3 3 3 3 3 3 3 3 3 3
+ nn_padding_conv: 1 1 1 1 1 1 1 1 1 1 1 1 1
+ nn_stride_conv: 1 1 1 1 1 1 1 1 1 1 1 1 1
+ nn_in_number_conv: 3 64 64 128 128 256 256 256 512 512 512 512 512
+ nn_out_number_conv: 64 64 128 128 256 256 256 512 512 512 512 512 512
+ nn_group_conv: 1 1 1 1 1 1 1 1 1 1 1 1 1
+ nn_bias_conv: 64 64 128 128 256 256 256 512 512 512 512 512 512
+ nn_in_data_size_pooling: 224 112 56 28 14
+ nn_channel_size_pooling: 2 2 2 2 2
+ nn_padding_pooling: 0 0 0 0 0
+ nn_stride_pooling: 2 2 2 2 2
+ nn_in_number_pooling: 64 128 256 512 512
+ nn_in_data_size_fc: 7 1 1
+ nn_in_number_fc: 512 4096 4096
+ nn_out_number_fc: 4096 4096 1000
+ nn_channel_size_fc: 7 1 1

codeGenerator/local_search.py

@@ -22,7 +22,7 @@ def conv_layer_perf(n, m, r, s, k, Tn, Tm, P_const, Tr, Tc):
     lat_read = math.ceil((min(Tn, n)/float(8))) * ((Tr-1)*s + k) * ((Tr-1)*s + k)
     # lat_read = 0
     if n == 3:
-        lat_com = Tr * Tc * math.ceil(k*k/2)
+        lat_com = Tr * Tc * math.ceil(k*k)
     else:
         lat_com = Tr * Tc * k * k
 

codeGenerator/tm_tn_multiAcc.py

@@ -69,7 +69,7 @@ def multiAcc_dse():
     """
     step 1: extract model from the original txt file with parameter no_include_fc / include_fc
     """
-    conv_N, conv_M, conv_r, conv_R, conv_K, conv_S, flag = model_extract('include_fc')
+    conv_N, conv_M, conv_r, conv_R, conv_K, conv_S, flag = model_extract('no_include_fc')
     OPs = gop_calculate(conv_N, conv_M, conv_R, conv_K)
     max_layerout = max_layer_dataout(conv_N, conv_M, conv_R, conv_K)