* Issues with old graphic cards is solved:

* CL sources should be saved as plain text (ANSI, not UTF). Some old OpenCL compilers return error in case of UTF files.
* Some graphic cards (Intel(R) HD Graphics 4400) cannot use doubles (no "cl_khr_fp64" extension). Therefore the CL sources were divided to float and double versions. The buttons "Run VBA performance test" and "Asynchronous execution" runs the "float" versions of CL sources because they can be executed at all kind of devices (hopefully).
* Old version of Inno setup was used to compile the installation. Latest Inno setup version created the installation that was falsely detected as virus.

Author: Excel-lent

C#/Installation script.iss

@@ -3,7 +3,7 @@ AppName=ClooWrapperVBA
 AppVerName=ClooWrapperVBA
 DefaultDirName={pf}\ClooWrapperVBA
 DefaultGroupName=ClooWrapperVBA
-Compression=lzma
+Compression=zip
 SolidCompression=yes
 SourceDir=.\
 PrivilegesRequired=poweruser
@@ -22,9 +22,11 @@ Name: "{app}\demo\cl"; Permissions: everyone-full
 Source: bin\ClooWrapperVBA.dll; DestDir: {app}; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;  
 Source: bin\ClooWrapperVBA_x64.dll; DestDir: {app}; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
 Source: bin\Cloo.dll; DestDir: {app}; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
-Source: ..\Excel\OpenCl v0.05.xlsm; DestDir: {app}\demo; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
-Source: ..\Excel\cl\Performance.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
-Source: ..\Excel\cl\MatrixMultiplication.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
+Source: ..\Excel\OpenCl example.xlsm; DestDir: {app}\demo; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
+Source: ..\Excel\cl\FloatPerformance.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
+Source: ..\Excel\cl\DoublePerformance.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
+Source: ..\Excel\cl\FloatMatrixMultiplication.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
+Source: ..\Excel\cl\DoubleMatrixMultiplication.cl; DestDir: {app}\demo\cl; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
 Source: ..\Excel\Configuration.vbs; DestDir: {app}\demo; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
 Source: bin\register.bat; DestDir: {app}; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;
 Source: bin\unregister.bat; DestDir: {app}; Flags: ignoreversion recursesubdirs overwritereadonly; Permissions: everyone-full;

Excel/Asynchronous.bas

@@ -10,7 +10,7 @@ Const THREAD_PRIORITY = 0
 Dim progDevices As Collection
 Dim currentTaskId&()
 
-Dim vecResp#()
+Dim vecResp!()
 Dim wsAsynchronous As Worksheet
 Dim globalWorkSize&(1), localWorkSize&(), globalWorkOffset&()
 Dim logLine%
@@ -26,7 +26,7 @@ Sub MainLoop()
     While Not allTasks_Completed
         For i = 1 To progDevices.Count
             If progDevices.Item(i).ProgramDevice.ExecutionCompleted Then
-                result = progDevices.Item(i).ProgramDevice.GetMemoryArgument_Double(0, vecResp)    ' Extract the results and do something with received data here.
+                result = progDevices.Item(i).ProgramDevice.GetMemoryArgument_Single(0, vecResp)    ' Extract the results and do something with received data here.
 
                 wsAsynchronous.Cells(logLine, 1) = "Task " & currentTaskId(i) & ", " & progDevices.Item(i).ProgramDevice.deviceType & _
                     progDevices.Item(i).DeviceId & ": completed"
@@ -37,7 +37,7 @@ Sub MainLoop()
                 ' Start new task
                 If startedTasks < MAX_TASKS Then
                     ReDim vecResp(UBound(vecResp))  ' Erase output vector.
-                    result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Double(0, vecResp)
+                    result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Single(0, vecResp)
 
                     ' If you want to use callbacks, than use function below
                     ' "CPU_Task_Completed" is a function that will obtain the callback.
@@ -83,14 +83,14 @@ Sub MainLoop()
 End Sub
 
 Sub RunAsynchronous()
-    Dim vecM1#(), vecM2#()
+    Dim vecM1!(), vecM2!()
     Dim vecQ&(1)
     Dim i&, j&, p&, q&, r&, nRows&
     Dim buildLogs$, sources$
 
     Set wsAsynchronous = ThisWorkbook.Worksheets("Asynchronous")
 
-    Open Application.ActiveWorkbook.Path & "\cl\MatrixMultiplication.cl" For Binary As #1
+    Open Application.ActiveWorkbook.Path & "\cl\FloatMatrixMultiplication.cl" For Binary As #1
     sources = Space$(LOF(1))
     Get #1, , sources
     Close #1
@@ -133,10 +133,10 @@ Sub RunAsynchronous()
 
     ReDim currentTaskId(progDevices.Count)
     For i = 1 To progDevices.Count
-        result = progDevices.Item(i).ProgramDevice.CreateKernel("DoubleMatrixMult")
-        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Double(0, vecResp)
-        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Double(1, vecM1)
-        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Double(2, vecM2)
+        result = progDevices.Item(i).ProgramDevice.CreateKernel("FloatMatrixMult")
+        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Single(0, vecResp)
+        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Single(1, vecM1)
+        result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Single(2, vecM2)
         result = progDevices.Item(i).ProgramDevice.SetMemoryArgument_Long(3, vecQ)
     Next i
 

Excel/HelloWorld.bas

@@ -6,10 +6,10 @@ Sub HelloWorld()
     Dim nRows%, currentRow&, nPlatforms&, nDevices&, i&, j&, result As Boolean
     Dim deviceType$, platformName$, platformVendor$, platformVersion$, deviceVendor$, deviceVersion$, driverVersion$, openCLCVersionString$
     Dim maxComputeUnits&, globalMemorySize#, maxClockFrequency#, maxMemoryAllocationSize#, deviceName$, sources$, cpuCounter&, gpuCounter&
-    Dim buildLogs$, platformId&, deviceId&, errorString$
+    Dim buildLogs$, platformId&, DeviceId&, errorString$
     Dim deviceAvailable As Boolean, compilerAvailable As Boolean
-    Dim m1#(1, 1), m2#(1, 1), vecM1#(), vecM2#(), vecQ&(0), vecResp#(3), globalWorkOffset&(), globalWorkSize&(1), localWorkSize&()
-    Dim p&, q&, r&, resp#()
+    Dim m1!(1, 1), m2!(1, 1), vecM1!(), vecM2!(), vecQ&(0), vecResp!(3), globalWorkOffset&(), globalWorkSize&(1), localWorkSize&()
+    Dim p&, q&, r&, resp!()
 
     Dim clooConfiguration As New ClooWrapperVBA.Configuration
     Dim progDevice As ClooWrapperVBA.ProgramDevice
@@ -76,7 +76,7 @@ Sub HelloWorld()
 
     ' Multiplication of two matrices.
     ' Read the OpenCL sources.
-    Open Application.ActiveWorkbook.Path & "\cl\MatrixMultiplication.cl" For Binary As #1
+    Open Application.ActiveWorkbook.Path & "\cl\FloatMatrixMultiplication.cl" For Binary As #1
     sources = Space$(LOF(1))
     Get #1, , sources
     Close #1
@@ -87,13 +87,13 @@ Sub HelloWorld()
         result = clooConfiguration.SetPlatform(platformId)
         cpuCounter = 0
         gpuCounter = 0
-        For deviceId = 0 To clooConfiguration.Platform.Devices - 1
-            result = clooConfiguration.Platform.SetDevice(deviceId)
+        For DeviceId = 0 To clooConfiguration.Platform.Devices - 1
+            result = clooConfiguration.Platform.SetDevice(DeviceId)
 
             If clooConfiguration.Platform.device.compilerAvailable Then
                 If clooConfiguration.Platform.device.deviceType = "CPU" Then
                     Set progDevice = New ClooWrapperVBA.ProgramDevice
-                    result = progDevice.Build(sources, "", platformId, deviceId, cpuCounter, buildLogs)
+                    result = progDevice.Build(sources, "", platformId, DeviceId, cpuCounter, buildLogs)
                     If result Then
                         Exit Do
                     Else
@@ -102,7 +102,7 @@ Sub HelloWorld()
                 End If
                 If clooConfiguration.Platform.device.deviceType = "GPU" Then
                     Set progDevice = New ClooWrapperVBA.ProgramDevice
-                    result = progDevice.Build(sources, "", platformId, deviceId, gpuCounter, buildLogs)
+                    result = progDevice.Build(sources, "", platformId, DeviceId, gpuCounter, buildLogs)
                     gpuCounter = gpuCounter + 1
                     If result Then
                         Exit Do
@@ -111,12 +111,12 @@ Sub HelloWorld()
                     End If
                 End If
             End If
-        Next deviceId
+        Next DeviceId
         platformId = platformId + 1
     Loop
 
     errorString = progDevice.errorString
-    result = progDevice.CreateKernel("DoubleMatrixMult")
+    result = progDevice.CreateKernel("FloatMatrixMult")
 
     ' Initialization of arrays:
     p = 2: q = 2: r = 2
@@ -125,28 +125,28 @@ Sub HelloWorld()
             m1(i, j) = wsHelloWorld.Cells(i + 1, j + 7)
         Next j
     Next i
-    vecM1 = MatrixToVector(m1, p, q)
+    vecM1 = MatrixToVectorSingle(m1, p, q)
     For i = 0 To q - 1
         For j = 0 To r - 1
             m2(i, j) = wsHelloWorld.Cells(i + 3, j + 7)
         Next j
     Next i
-    vecM2 = MatrixToVector(m2, q, r)
+    vecM2 = MatrixToVectorSingle(m2, q, r)
     vecQ(0) = q
 
-    result = progDevice.SetMemoryArgument_Double(0, vecResp)
-    result = progDevice.SetMemoryArgument_Double(1, vecM1)
-    result = progDevice.SetMemoryArgument_Double(2, vecM2)
+    result = progDevice.SetMemoryArgument_Single(0, vecResp)
+    result = progDevice.SetMemoryArgument_Single(1, vecM1)
+    result = progDevice.SetMemoryArgument_Single(2, vecM2)
     result = progDevice.SetMemoryArgument_Long(3, vecQ)
 
     globalWorkSize(0) = p
     globalWorkSize(1) = r
 
     result = progDevice.ExecuteSync(globalWorkOffset, globalWorkSize, localWorkSize)
 
-    result = progDevice.GetMemoryArgument_Double(0, vecResp)
+    result = progDevice.GetMemoryArgument_Single(0, vecResp)
 
-    resp = VectorToMatrix(vecResp, p, r)
+    resp = VectorToMatrixSingle(vecResp, p, r)
 
     For i = 0 To p - 1
         For j = 0 To r - 1

Excel/Helpers.bas

@@ -37,12 +37,12 @@ Function CreateDeviceCollection(sources$)
 
             If clooConfiguration.Platform.device.deviceType = "CPU" Then
                 result = progDevice.ProgramDevice.Build(sources, "", i - 1, j - 1, cpuCounter, buildLogs)
-                progDevice.deviceId = cpuCounter
+                progDevice.DeviceId = cpuCounter
                 progDevice.deviceType = "CPU"
                 If result = True Then cpuCounter = cpuCounter + 1
             ElseIf clooConfiguration.Platform.device.deviceType = "GPU" Then
                 result = progDevice.ProgramDevice.Build(sources, "", i - 1, j - 1, gpuCounter, buildLogs)
-                progDevice.deviceId = gpuCounter
+                progDevice.DeviceId = gpuCounter
                 progDevice.deviceType = "GPU"
                 If result = True Then gpuCounter = gpuCounter + 1
             Else
@@ -62,7 +62,37 @@ Function CreateDeviceCollection(sources$)
     End If
 End Function
 
-Function MatrixToVector(m() As Double, maxi As Long, maxj As Long) As Double()
+Function MatrixToVectorSingle(m() As Single, maxi As Long, maxj As Long) As Single()
+    Dim v() As Single
+    Dim i&, j&
+    
+    ReDim v(maxi * maxj - 1)
+    
+    For i = 0 To maxi - 1
+        For j = 0 To maxj - 1
+            v(i + maxi * j) = m(i, j)
+        Next j
+    Next i
+    
+    MatrixToVectorSingle = v
+End Function
+
+Function VectorToMatrixSingle(v() As Single, maxi As Long, maxj As Long) As Single()
+    Dim i&, j&
+    Dim m() As Single
+    
+    ReDim m(maxi - 1, maxj - 1)
+    
+    For i = 0 To maxi - 1
+        For j = 0 To maxj - 1
+            m(i, j) = v(i + maxi * j)
+        Next j
+    Next i
+    
+    VectorToMatrixSingle = m
+End Function
+
+Function MatrixToVectorDouble(m() As Double, maxi As Long, maxj As Long) As Double()
     Dim v() As Double
     Dim i&, j&
 
@@ -74,10 +104,10 @@ Function MatrixToVector(m() As Double, maxi As Long, maxj As Long) As Double()
         Next j
     Next i
 
-    MatrixToVector = v
+    MatrixToVectorDouble = v
 End Function
 
-Function VectorToMatrix(v() As Double, maxi As Long, maxj As Long) As Double()
+Function VectorToMatrixDouble(v() As Double, maxi As Long, maxj As Long) As Double()
     Dim i&, j&
     Dim m() As Double
 
@@ -89,5 +119,5 @@ Function VectorToMatrix(v() As Double, maxi As Long, maxj As Long) As Double()
         Next j
     Next i
 
-    VectorToMatrix = m
+    VectorToMatrixDouble = m
 End Function