allocate memory in block rather than buffer for avoiding too small buffer size

source/network/XBackwardFunc.cpp

@@ -31,10 +31,16 @@ namespace nts{
 /* compute dE/dx of a node */
 /* compute dE/dx of a node */
 void XFuncGrad::MakeGrad(XTensor * node, bool isEfficient)
 void XFuncGrad::MakeGrad(XTensor * node, bool isEfficient)
 {
 {
+    if (!isEfficient) {
+        CheckNTErrors(node->grad != NULL, "No gradient found!");
+    }
+    else {
+        CheckNTErrors(!node->isGrad || node->grad != NULL, "No gradient found!");
+    }
+
     XLink &income = node->income;
     XLink &income = node->income;
     int operID = income.typeID;
     int operID = income.typeID;
 
 
-    CheckNTErrors(node->grad != NULL, "No gradient found!");
     CheckNTErrors(income.tailNum == 1, "Too many input tensors for the function!");
     CheckNTErrors(income.tailNum == 1, "Too many input tensors for the function!");
 
 
     XTensor * input = income.tails[0];
     XTensor * input = income.tails[0];
@@ -45,7 +51,8 @@ void XFuncGrad::MakeGrad(XTensor * node, bool isEfficient)
 
 
         XTensor * dedx = input->grad;
         XTensor * dedx = input->grad;
         XTensor * dedy = output->grad;
         XTensor * dedy = output->grad;
-        XTensor * tmp = NewTensorBufV2(output, output->devID, output->mem);
+        //XTensor * tmp = NewTensorBufV2(output, output->devID, output->mem);
+        XTensor * tmp = NewTensor(output);
 
 
         if (operID == FUNC_HARDTANH)
         if (operID == FUNC_HARDTANH)
             _HardTanHBackward(output, input, dedy, tmp);
             _HardTanHBackward(output, input, dedy, tmp);
@@ -70,7 +77,8 @@ void XFuncGrad::MakeGrad(XTensor * node, bool isEfficient)
         }
         }
 
 
         _SumMe(dedx, tmp);
         _SumMe(dedx, tmp);
-        DelTensorBuf(tmp);
+        //DelTensorBuf(tmp);
+        DelTensor(tmp);
     }
     }
 
 
     node->visitMark = NODE_FINISHED;
     node->visitMark = NODE_FINISHED;

source/network/XBackwardLoss.cpp

@@ -58,7 +58,9 @@ void XLossGrad::MakeGrad(XTensor * node, bool isEfficient)
 
 
         gold = income.tails[1];
         gold = income.tails[1];
 
 
-        XTensor* tmp = NewTensorBufV2(output, output->devID, output->mem);
+        //XTensor * tmp = NewTensorBufV2(output, output->devID, output->mem);
+        XTensor* tmp = NewTensor(output);
+
         if (operID == LOSS_CROSSENTROPY) {
         if (operID == LOSS_CROSSENTROPY) {
             if (income.tailNum == 3)
             if (income.tailNum == 3)
                 padding = income.tails[2];
                 padding = income.tails[2];
@@ -70,7 +72,8 @@ void XLossGrad::MakeGrad(XTensor * node, bool isEfficient)
         else {
         else {
             ShowNTErrors("Wrong activation function type!");
             ShowNTErrors("Wrong activation function type!");
         }
         }
-        DelTensorBuf(tmp);
+        //DelTensorBuf(tmp);
+        DelTensor(tmp);
     }
     }
 
 
     node->visitMark = NODE_FINISHED;
     node->visitMark = NODE_FINISHED;