Revert "passing through reductionOp. Not yet compiling."

This reverts commit 84c9b0caa0.

Source/Math/GPUTensor.cu

@@ -268,7 +268,7 @@ struct TensorOps
 //#define ReduceElemType double
 #define ReduceElemType ElemType // (note: we could use 'double' here, but that would cause problems with CUDA cards that don't support double)
 
-template <class ElemType, ElementWiseOperator reductionOp, C_size_t N, C_int M, C_int m>
+template <class ElemType, C_size_t N, C_int M, C_int m>
 struct TensorOpReduce
 {
     // this version for m >= 0
@@ -277,7 +277,7 @@ struct TensorOpReduce
     {
         // start with index 0
         // We may use 'double' since we are memory-bound anyway.
-        ReduceElemType aggregate = TensorOpReduce<ElemType, reductionOp, N, M, m - 1>::Compute(pointers, op, reducingOpDims, reducingStrides);
+        ReduceElemType aggregate = TensorOpReduce<ElemType, N, M, m - 1>::Compute(pointers, op, reducingOpDims, reducingStrides);
         // apply this index to the pointers
         C_size_t dim = reducingOpDims[m];
         for (C_size_t k = 1 /*done with k=0 already*/; k < dim; k++)
@@ -285,21 +285,8 @@ struct TensorOpReduce
             // bump the pointers
             for (C_size_t i = 0; i < N - 1; i++) // N-1 because output is not used here
                 pointers[i] += reducingStrides(i, (C_size_t) m);
-            ElemType val = TensorOpReduce<ElemType, reductionOp, N, M, m - 1>::Compute(pointers, op, reducingOpDims, reducingStrides);
-
-            switch (reductionOp)
-            {
-            case ElementWiseOperator::opSum:
-                aggregate += val;
-                break;
-            case ElementWiseOperator::opMax:
-                aggregate = std::max<ReduceElemType>(aggregate, val);
-                break;
-            case ElementWiseOperator::opMin:
-                aggregate = std::max<ReduceElemType>(aggregate, val);
-                break;
-            }
-
+            ElemType val = TensorOpReduce<ElemType, N, M, m - 1>::Compute(pointers, op, reducingOpDims, reducingStrides);
+            aggregate += val;
         }
         return (ElemType) aggregate;
     }
@@ -308,7 +295,7 @@ struct TensorOpReduce
 // this one terminates the template recursion over reduction dimensions
 // The pointers are pointing to the input element.
 template <class ElemType, C_size_t N, C_int M>
-struct TensorOpReduce<ElemType, ElementWiseOperator reductionOp, N, M, /*m=*/-1>
+struct TensorOpReduce<ElemType, N, M, /*m=*/-1>
 {
     // this version for m = -1
     // the pointers are pointing to the right location(s) to take the operation over