For all the wprint related stuff, we should avoid use %s because wstring is the internal representation (no conversion needed). Replaced it with %ls.

Demos/Simple/Simple.config

@@ -2,7 +2,7 @@
 command=Simple_Demo:Simple_Demo_Output
 
 # deviceId=-1 for CPU, >=0 for GPU devices
-DeviceNumber=0
+DeviceNumber=-1
 #stderr=Demo
 
 precision=float
@@ -45,14 +45,14 @@ Simple_Demo=[
         learningRatesPerMB=0.5:0.2*20:0.1
         momentumPerMB=0.9
         dropoutRate=0.0
-        maxEpochs=500
+        maxEpochs=50
     ]
        
     # Parameter values for the reader
     reader=[
       # reader to use
       readerType=UCIFastReader
-      file=SimpleDataTrain.txt
+      file=../Demos/Simple/SimpleDataTrain.txt
 
       miniBatchMode=Partial
       randomize=Auto
@@ -67,7 +67,7 @@ Simple_Demo=[
 	start=2      # Skip two elements
         dim=1        # One label dimension
         labelDim=2   # Two labels possible
-        labelMappingFile=SimpleMapping.txt
+        labelMappingFile=../Demos/Simple/SimpleMapping.txt
       ]
     ]
 ]
@@ -84,7 +84,7 @@ Simple_Demo_Output=[
     reader=[
       # reader to use
       readerType=UCIFastReader
-      file=SimpleDataTest.txt
+      file=../Demos/Simple/SimpleDataTest.txt
       features=[
           dim=2
 	  start=0
@@ -93,7 +93,7 @@ Simple_Demo_Output=[
 	start=2
         dim=1
         labelDim=2
-        labelMappingFile=SimpleMapping.txt
+        labelMappingFile=../Demos/Simple/SimpleMapping.txt
       ]
     ]
     outputPath=SimpleOutput    # Dump output as text

MachineLearning/cn/ComputationNetwork.h

@@ -172,7 +172,7 @@ private:	// [erw] added for Toplological Plot only
             }
 			wstring str = style; 
 			for (auto x : specialNodes){
-				str = str + msra::strfun::wstrprintf(L"\"%S\" ", x->GetName().c_str());
+				str = str + msra::strfun::wstrprintf(L"\"%ls\" ", x->GetName().c_str());
 			}
 			return str + L"; \n";
 		}
@@ -243,10 +243,10 @@ public:
 				line.clear(); 
                 size_t nrows = x->FunctionValues().GetNumRows();
                 size_t ncols = x->FunctionValues().GetNumCols();
-				line = msra::strfun::wstrprintf(L" \"%S\" [ label = \"%S [%d,%d]\\n%S\" ] ;\n", 
+				line = msra::strfun::wstrprintf(L" \"%ls\" [ label = \"%ls [%d,%d]\\n%ls\" ] ;\n", 
                     x->GetName().c_str(), x->GetName().c_str(), nrows, ncols,  x->OperationName().c_str());
-                /*fprintf(stderr, "%S\n", line.c_str());
-                line += L" \"%s\" [ label = \"%s [%d,%d]\\n%s\" ] ;\n";
+                //fprintf(stderr, "%ls\n", line.c_str());
+                /*  line += L" \"%s\" [ label = \"%s [%d,%d]\\n%s\" ] ;\n";
                 line += x->GetName().c_str();
                 line += x->GetName().c_str();
                 line += L" \"%s\" [ label = \"%s [%d,%d]\\n%s\" ] ;\n";*/
@@ -263,7 +263,7 @@ public:
 			line.clear(); 
 			for (auto x : m_features)
 			{
-				line = line + msra::strfun::wstrprintf(L"\"%S\" ", x->GetName().c_str());
+				line = line + msra::strfun::wstrprintf(L"\"%ls\" ", x->GetName().c_str());
 			}
 			fstream << line << L"\n}\n"; 
 			// subgraph eval/output/criteria
@@ -272,15 +272,15 @@ public:
 			line.clear(); 
 			for (auto x : m_finalCriteria)
 			{
-				line = line + msra::strfun::wstrprintf(L"\"%S\" ", x->GetName().c_str());
+				line = line + msra::strfun::wstrprintf(L"\"%ls\" ", x->GetName().c_str());
 			}
 			for (auto x : m_outputNodes)
 			{
-				line = line + msra::strfun::wstrprintf(L"\"%S\" ", x->GetName().c_str());
+				line = line + msra::strfun::wstrprintf(L"\"%ls\" ", x->GetName().c_str());
 			}
 			for (auto x : m_evalNodes)
 			{
-				line = line + msra::strfun::wstrprintf(L"\"%S\" ", x->GetName().c_str());
+				line = line + msra::strfun::wstrprintf(L"\"%ls\" ", x->GetName().c_str());
 			}
 			fstream << line << L"\n}\n"; 
 
@@ -302,14 +302,14 @@ public:
 					// create a dummy node 
 					ComputationNodePtr delayedNode = des;
 					wstring dummyName = des->GetName() + L".dummy";
-					wstring out = msra::strfun::wstrprintf(L"node [ shape = box3d  , color = lightgray, style = \"filled\" , label = \"%S\" ] ; \"%S\"\n",
+					wstring out = msra::strfun::wstrprintf(L"node [ shape = box3d  , color = lightgray, style = \"filled\" , label = \"%ls\" ] ; \"%ls\"\n",
 						(delayedNode->GetName() + L"\\n(delayed)").c_str(), dummyName.c_str());
 					line = out; 
-					line += msra::strfun::wstrprintf(L"\"%S\" -> \"%S\" ; \n", dummyName.c_str(), srcname.c_str());
+					line += msra::strfun::wstrprintf(L"\"%ls\" -> \"%ls\" ; \n", dummyName.c_str(), srcname.c_str());
 				}
 				else
 				{
-					line = msra::strfun::wstrprintf(L"\"%S\" -> \"%S\" ; \n", desname.c_str(), srcname.c_str());
+					line = msra::strfun::wstrprintf(L"\"%ls\" -> \"%ls\" ; \n", desname.c_str(), srcname.c_str());
 				}
 
 

Makefile_kaldi2.gpu

@@ -35,7 +35,7 @@ BUILDTYPE = release
 #MATHLIB = acml
 MATHLIB = mkl
 # modify relevant path below for your system
-MKL_PATH = /usr/users/chiaying/intel/composer_xe_2013.2.146
+MKL_PATH = /usr/users/yzhang87/tools/composer_xe_2015.2.164
 ACML_PATH = /usr/users/yzhang87/code/acml/gfortran64
 #######