renamed DataTensor.h to TensorShape.h. No code changes

Source/CNTK/CNTK.vcxproj

@@ -158,7 +158,7 @@
     <ClInclude Include="..\Common\Include\Basics.h" />
     <ClInclude Include="..\Common\Include\BestGpu.h" />
     <ClInclude Include="..\Common\Include\DataReader.h" />
-    <ClInclude Include="..\Common\Include\DataTensor.h" />
+    <ClInclude Include="..\Common\Include\TensorShape.h" />
     <ClInclude Include="..\Common\Include\DataWriter.h" />
     <ClInclude Include="..\Common\Include\File.h" />
     <ClInclude Include="..\Common\Include\fileutil.h" />

Source/CNTK/CNTK.vcxproj.filters

@@ -133,7 +133,7 @@
     <ClInclude Include="..\Common\Include\Sequences.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
-    <ClInclude Include="..\Common\Include\DataTensor.h">
+    <ClInclude Include="..\Common\Include\TensorShape.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
     <ClInclude Include="..\Common\Include\ProgressTracing.h">

Source/CNTK/SynchronousExecutionEngine.cpp

@@ -15,7 +15,7 @@
 #include "ConvolutionalNodes.h"
 #include "NonlinearityNodes.h"
 #include "ReshapingNodes.h"
-#include "DataTensor.h"
+#include "TensorShape.h"
 
 namespace Microsoft { namespace MSR { namespace CNTK {
 

Source/Common/Include/TensorShape.h

@@ -1,4 +1,4 @@
-// DataTensor.h -- tensor descriptor that describes the inner structure of data vectors
+// TensorShape.h -- tensor descriptor that describes the inner structure of data vectors
 //
 // <copyright file="Sequences.h" company="Microsoft">
 //     Copyright (c) Microsoft Corporation.  All rights reserved.

Source/ComputationNetworkLib/ComputationNetworkBuilder.h

@@ -9,7 +9,7 @@
 #include "ComputationNetwork.h"
 #include "TrainingCriterionNodes.h" // for NCEEvalMode
 #include "ScriptableObjects.h"
-#include "DataTensor.h"
+#include "TensorShape.h"
 #include <string>
 
 namespace Microsoft { namespace MSR { namespace CNTK {

Source/ComputationNetworkLib/ComputationNetworkLib.vcxproj

@@ -155,7 +155,7 @@
     <ClInclude Include="..\Common\Include\Basics.h" />
     <ClInclude Include="..\Common\Include\BestGpu.h" />
     <ClInclude Include="..\Common\Include\Config.h" />
-    <ClInclude Include="..\Common\Include\DataTensor.h" />
+    <ClInclude Include="..\Common\Include\TensorShape.h" />
     <ClInclude Include="..\Common\Include\File.h" />
     <ClInclude Include="..\Common\Include\fileutil.h" />
     <ClInclude Include="..\Common\Include\Platform.h" />

Source/ComputationNetworkLib/ComputationNetworkLib.vcxproj.filters

@@ -117,7 +117,7 @@
     <ClInclude Include="EsotericNodes.h">
       <Filter>Nodes</Filter>
     </ClInclude>
-    <ClInclude Include="..\Common\Include\DataTensor.h">
+    <ClInclude Include="..\Common\Include\TensorShape.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
     <ClInclude Include="..\Common\Include\Config.h">

Source/ComputationNetworkLib/ComputationNode.cpp

@@ -9,7 +9,7 @@
 #include "ComputationNode.h"
 #include "InputAndParamNodes.h"
 #include "ComputationNetworkBuilder.h"  // TODO: We should only pull in NewComputationNodeFromConfig(). Nodes should not know about network at large.
-#include "DataTensor.h"
+#include "TensorShape.h"
 
 #ifndef let
 #define let const auto

Source/ComputationNetworkLib/ComputationNode.h

@@ -10,7 +10,7 @@
 #include "TensorView.h"
 #include "ScriptableObjects.h"
 #include "Sequences.h"
-#include "DataTensor.h"
+#include "TensorShape.h"
 #include "MatrixPool.h"
 
 #include <unordered_set>

Source/Math/ConvolutionEngine.h

@@ -18,7 +18,7 @@
 #endif
 
 #include "Matrix.h"
-#include "DataTensor.h" // for ImageLayoutKind
+#include "TensorShape.h" // for ImageLayoutKind
 
 namespace Microsoft { namespace MSR { namespace CNTK {
 

Source/Math/GPUMatrix.h

@@ -9,7 +9,7 @@
 #include "File.h"
 #include "Helpers.h"
 #include "CommonMatrix.h"
-#include "DataTensor.h" // only for SmallVector; I was hoping to keep this out
+#include "TensorShape.h" // only for SmallVector; I was hoping to keep this out
 #include "DebugUtil.h"
 #include "BestGpu.h"    // for CPUONLY macro
 #include "ConcStack.h"

Source/Math/GPUTensor.cu

@@ -378,6 +378,7 @@ namespace Microsoft { namespace MSR { namespace CNTK {
             {
                 if (tid < i && tid + i < tids) accumulators[tid] += accumulators[tid + i];
                 if (0 + i < tids) __syncthreads();    // sync if condition true for at least one thread
+                // TODO: use volatile* and then we can skip the __syncthreads() for the last 32 values
             }
 
             // now set final value to output coordinate

Source/Math/GPUTensor.h

@@ -6,7 +6,7 @@
 
 #pragma once
 #include "CommonMatrix.h"
-#include "DataTensor.h" // only for SmallVector; I was hoping to keep this out
+#include "TensorShape.h" // only for SmallVector; I was hoping to keep this out
 #include "GPUMatrixCUDAKernels.cuh"
 #include <array>
 

Source/Math/Math.vcxproj

@@ -156,7 +156,7 @@
     </ProjectReference>
   </ItemDefinitionGroup>
   <ItemGroup>
-    <ClInclude Include="..\Common\Include\DataTensor.h" />
+    <ClInclude Include="..\Common\Include\TensorShape.h" />
     <ClInclude Include="..\Common\Include\File.h" />
     <ClInclude Include="..\Common\Include\fileutil.h" />
     <ClInclude Include="..\Common\Include\DebugUtil.h" />

Source/Math/Math.vcxproj.filters

@@ -76,7 +76,7 @@
     <ClInclude Include="TensorOps.h">
       <Filter>Tensors</Filter>
     </ClInclude>
-    <ClInclude Include="..\Common\Include\DataTensor.h">
+    <ClInclude Include="..\Common\Include\TensorShape.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
     <ClInclude Include="Helpers.h">

Source/Math/Matrix.h

@@ -13,7 +13,7 @@
 #include "Basics.h"
 #include "File.h"
 #include "CommonMatrix.h"
-#include "DataTensor.h" // only for SmallVector; I was hoping to keep this out
+#include "TensorShape.h" // only for SmallVector; I was hoping to keep this out
 #include <limits.h>
 #include <memory>       // for shared_ptr
 #include <array>

Source/Math/NoGPU.cpp

@@ -13,7 +13,7 @@
 #include "GPUSparseMatrix.h"
 #include "MatrixQuantizerGPU.h"
 #include "CuDnnConvolutionEngine.h"
-#include "DataTensor.h"
+#include "TensorShape.h"
 
 #pragma warning (disable: 4100) // unreferenced formal parameter, which is OK since all functions in here are dummies; disabling this allows to copy-paste prototypes here when we add new functions
 #pragma warning (disable: 4702) // unreachable code, which we get from the NOT_IMPLEMENTED macro which is OK

Source/Math/TensorView.h

@@ -10,7 +10,7 @@
 
 #include "Basics.h"
 #include "Matrix.h"
-#include "DataTensor.h"
+#include "TensorShape.h"
 
 #pragma warning (push)
 #pragma warning (disable: 4251) // needs to have dll-interface to be used by clients of... caused by TensorView::m_shape which is only private. We use the same compiler everywhere.

Source/SGDLib/SGDLib.vcxproj

@@ -156,7 +156,7 @@
     <ClInclude Include="..\Common\Include\BestGpu.h" />
     <ClInclude Include="..\Common\Include\Config.h" />
     <ClInclude Include="..\Common\Include\DataReader.h" />
-    <ClInclude Include="..\Common\Include\DataTensor.h" />
+    <ClInclude Include="..\Common\Include\TensorShape.h" />
     <ClInclude Include="..\Common\Include\DataWriter.h" />
     <ClInclude Include="..\Common\Include\File.h" />
     <ClInclude Include="..\Common\Include\fileutil.h" />

Source/SGDLib/SGDLib.vcxproj.filters

@@ -141,7 +141,7 @@
     <ClInclude Include="..\Common\Include\Sequences.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
-    <ClInclude Include="..\Common\Include\DataTensor.h">
+    <ClInclude Include="..\Common\Include\TensorShape.h">
       <Filter>Common\Include</Filter>
     </ClInclude>
     <ClInclude Include="..\Common\Include\Config.h">