Added a more verbose version of getting the stats for the devices

src/MATLAB/+ImProc/@Cuda/Cuda.m

@@ -7,6 +7,7 @@ methods (Static)
     imageOut = ApplyPolyTransformation(imageIn,a,b,c,min,max,device)
     imageOut = ContrastEnhancement(imageIn,sigma,MedianNeighborhood,device)
     [numCudaDevices,memoryStats] = DeviceCount()
+    deviceStatsArray = DeviceStats()
     imageOut = GaussianFilter(imageIn,sigma,device)
     histogram = Histogram(imageIn,numBins,min,max,device)
     imageOut = ImagePow(imageIn,power,device)

src/MATLAB/+ImProc/@Cuda/DeviceCount.m

-% DeviceCount - This will return statistics on the Cuda devices available.
+% DeviceCount - This will return the number of Cuda devices available, and their memory.
 %    [numCudaDevices,memoryStats] = ImProc.Cuda.DeviceCount()
 %    	NumCudaDevices -- this is the number of Cuda devices available.
-%    	MemoryStats -- this is an array of structures where each entery corresponds to a Cuda device.
+%    	MemoryStats -- this is an array of structures where each entry corresponds to a Cuda device.
 %    		The memory structure contains the total memory on the device and the memory available for a Cuda call.
 function [numCudaDevices,memoryStats] = DeviceCount()
     [numCudaDevices,memoryStats] = ImProc.Cuda.Mex('DeviceCount');

src/MATLAB/+ImProc/@Cuda/DeviceStats.m

+% DeviceStats - This will return the statistics of each Cuda capable device installed.
+%    deviceStatsArray = ImProc.Cuda.DeviceStats()
+%    	DeviceStatsArray -- this is an array of structs, one struct per device.
+%    		The struct has these fields: name, major, minor, constMem, sharedMem, totalMem, tccDriver, mpCount, threadsPerMP, warpSize, maxThreads.
+function deviceStatsArray = DeviceStats()
+    [deviceStatsArray] = ImProc.Cuda.Mex('DeviceStats');
+end

src/MATLAB/+ImProc/BuildScript.m

@@ -3,7 +3,8 @@
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 excludeList = ...
     {'Cuda.m';
-     'DeviceCount.m'};
+     'DeviceCount.m';
+     'DeviceStats.m'};
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 
 % remember curent path

src/MATLAB/+ImProc/DeviceStats.m

+% DeviceStats - This will return the statistics of each Cuda capable device installed.
+%    deviceStatsArray = ImProc.Devic);
+%    	DeviceStatsArray -- this is an array of structs, one struct per device.
+%    		The struct has these fields: name, major, minor, constMem, sharedMem, totalMem, tccDriver, mpCount, threadsPerMP, warpSize, maxThreads.
+function deviceStatsArray = DeviceStats()
+end
+

src/c/Cuda/CWrappers.cu

@@ -2,6 +2,7 @@
 #include "CudaAdd.cuh"
 #include "CudaContrastEnhancement.cuh"
 #include "CudaDeviceCount.cuh"
+#include "CudaDeviceStats.h"
 #include "CudaHistogram.cuh"
 #include "CudaImageCopy.cuh"
 #include "CudaGaussianFilter.cuh"
@@ -181,6 +182,11 @@ void clearDevice()
  }
 
 
+ int deviceStats(DevStats** stats)
+ {
+	 return cDeviceStats(stats);
+ }
+
  size_t* histogram(const unsigned char* imageIn, Vec<size_t> dims, unsigned int arraySize, unsigned char minVal/*=std::numeric_limits<unsigned char>::lowest()*/, unsigned char maxVal/*=std::numeric_limits<unsigned char>::max()*/, int device/*=0*/)
 {
 	return cCalculateHistogram(imageIn,dims,arraySize,minVal,maxVal,device);

src/c/Cuda/CWrappers.h

 #pragma once
 #include "Vec.h"
+#include "CudaDeviceStats.h"
 #include <limits>
 
 #ifdef IMAGE_PROCESSOR_DLL
@@ -47,6 +48,7 @@ IMAGE_PROCESSOR_API float* contrastEnhancement(const float* imageIn, Vec<size_t>
 IMAGE_PROCESSOR_API double* contrastEnhancement(const double* imageIn, Vec<size_t> dims, Vec<float> sigmas, Vec<size_t> neighborhood, double** imageOut = NULL, int device = 0);
 
 IMAGE_PROCESSOR_API int deviceCount();
+IMAGE_PROCESSOR_API int deviceStats(DevStats** stats);
 
 IMAGE_PROCESSOR_API size_t* histogram(const unsigned char* imageIn, Vec<size_t> dims, unsigned int arraySize, unsigned char minVal = std::numeric_limits<unsigned char>::lowest(), unsigned char maxVal = std::numeric_limits<unsigned char>::max(), int device = 0);
 IMAGE_PROCESSOR_API size_t* histogram(const unsigned short* imageIn, Vec<size_t> dims, unsigned int arraySize, unsigned short minVal = std::numeric_limits<unsigned short>::lowest(), unsigned short maxVal = std::numeric_limits<unsigned short>::max(), int device = 0);

src/c/Cuda/CudaDeviceStats.cpp

+#include "CudaDeviceStats.h"
+#include <cuda_runtime.h>
+
+int cDeviceStats(DevStats** stats)
+{
+    int cnt = 0;
+    cudaGetDeviceCount(&cnt);
+
+    *stats = new DevStats[cnt];
+
+    for(int device = 0; device<cnt; ++device)
+    {
+        cudaDeviceProp props;
+        cudaGetDeviceProperties(&props, device);
+
+        DevStats* curStats = (*stats)+device;
+        curStats->name = props.name;
+        curStats->major = props.major;
+        curStats->minor = props.minor;
+
+        curStats->constMem = props.totalConstMem;
+        curStats->sharedMem = props.sharedMemPerBlock;
+        curStats->totalMem = props.totalGlobalMem;
+
+        curStats->tccDriver = props.tccDriver>0;
+        curStats->mpCount = props.multiProcessorCount;
+        curStats->threadsPerMP = props.maxThreadsPerMultiProcessor;
+
+        curStats->warpSize = props.warpSize;
+        curStats->maxThreads = props.maxThreadsPerBlock;
+    }
+
+    return cnt;
+}
\ No newline at end of file

src/c/Cuda/CudaDeviceStats.h

+#pragma once
+#include <string>
+
+struct DevStats
+{
+    std::string name;
+    int major;
+    int minor;
+
+    size_t constMem;
+    size_t sharedMem;
+    size_t totalMem;
+
+    bool tccDriver;
+    int mpCount;
+    int threadsPerMP;
+
+    int warpSize;
+    int maxThreads;
+};
+
+int cDeviceStats(DevStats** stats);

src/c/CudaImageProcessor_dll.vcxproj

@@ -26,6 +26,7 @@
     <ClInclude Include="Cuda\CudaConvertType.cuh" />
     <ClInclude Include="Cuda\CudaDeviceCount.cuh" />
     <ClInclude Include="Cuda\CudaDeviceImages.cuh" />
+    <ClInclude Include="Cuda\CudaDeviceStats.h" />
     <ClInclude Include="Cuda\CudaGaussianFilter.cuh" />
     <ClInclude Include="Cuda\CudaGetMinMax.cuh" />
     <ClInclude Include="Cuda\CudaGetROI.cuh" />
@@ -67,6 +68,9 @@
     <ClCompile Include="Cuda\CHelpers.cpp" />
   </ItemGroup>
   <ItemGroup>
+    <ClCompile Include="Cuda\CudaDeviceStats.cpp">
+      <FileType>Document</FileType>
+    </ClCompile>
     <CudaCompile Include="Cuda\CudaUtilities.cu" />
     <CudaCompile Include="Cuda\CWrappers.cu" />
     <CudaCompile Include="Cuda\ImageChunk.cu" />

src/c/CudaImageProcessor_dll.vcxproj.filters

@@ -140,11 +140,17 @@
     <ClInclude Include="Cuda\CudaClamp.cuh">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="Cuda\CudaDeviceStats.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="Cuda\CHelpers.cpp">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="Cuda\CudaDeviceStats.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <CudaCompile Include="Cuda\CudaUtilities.cu">

src/c/CudaImageProcessor_lib.vcxproj

@@ -25,6 +25,7 @@
     <ClInclude Include="Cuda\CudaConvertType.cuh" />
     <ClInclude Include="Cuda\CudaDeviceCount.cuh" />
     <ClInclude Include="Cuda\CudaDeviceImages.cuh" />
+    <ClInclude Include="Cuda\CudaDeviceStats.h" />
     <ClInclude Include="Cuda\CudaGaussianFilter.cuh" />
     <ClInclude Include="Cuda\CudaGetMinMax.cuh" />
     <ClInclude Include="Cuda\CudaGetROI.cuh" />
@@ -62,6 +63,7 @@
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="Cuda\CHelpers.cpp" />
+    <ClCompile Include="Cuda\CudaDeviceStats.cpp" />
   </ItemGroup>
   <ItemGroup>
     <CudaCompile Include="Cuda\CudaUtilities.cu" />

src/c/CudaImageProcessor_lib.vcxproj.filters

@@ -131,11 +131,17 @@
     <ClInclude Include="Cuda\CudaStdFilter.cuh">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="Cuda\CudaDeviceStats.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="Cuda\CHelpers.cpp">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="Cuda\CudaDeviceStats.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <CudaCompile Include="Cuda\CudaUtilities.cu">

src/c/CudaMex.vcxproj

@@ -104,6 +104,7 @@ copy $(OutDir)CudaMex.dll "$(ProjectDir)Mex.mexw64"</Command>
     <ClCompile Include="Mex\MexCommand.cpp" />
     <ClCompile Include="Mex\MexContrastEnhancement.cpp" />
     <ClCompile Include="Mex\MexDeviceCount.cpp" />
+    <ClCompile Include="Mex\MexDeviceStats.cpp" />
     <ClCompile Include="Mex\MexGaussianFilter.cpp" />
     <ClCompile Include="Mex\MexHistogram.cpp" />
     <ClCompile Include="Mex\MexImagePow.cpp" />

src/c/CudaMex.vcxproj.filters

@@ -140,6 +140,9 @@
     <ClCompile Include="Mex\MexCommand.cpp">
       <Filter>Source Files</Filter>
     </ClCompile>
+    <ClCompile Include="Mex\MexDeviceStats.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <None Include="Mex\CudaMex.def">

src/c/Mex/CommandList.h

@@ -43,6 +43,7 @@ DEF_MEX_COMMAND(AddImageWith)
 DEF_MEX_COMMAND(ApplyPolyTransformation)
 DEF_MEX_COMMAND(ContrastEnhancement)
 DEF_MEX_COMMAND(DeviceCount)
+DEF_MEX_COMMAND(DeviceStats)
 DEF_MEX_COMMAND(GaussianFilter)
 DEF_MEX_COMMAND(Histogram)
 DEF_MEX_COMMAND(ImagePow)

src/c/Mex/MexDeviceCount.cpp

@@ -42,9 +42,9 @@ void MexDeviceCount::usage(std::vector<std::string>& outArgs,std::vector<std::st
 
 void MexDeviceCount::help(std::vector<std::string>& helpLines) const
 {
-	helpLines.push_back("This will return statistics on the Cuda devices available.");
+	helpLines.push_back("This will return the number of Cuda devices available, and their memory.");
 
 	helpLines.push_back("\tNumCudaDevices -- this is the number of Cuda devices available.");
-	helpLines.push_back("\tMemoryStats -- this is an array of structures where each entery corresponds to a Cuda device.");
+	helpLines.push_back("\tMemoryStats -- this is an array of structures where each entry corresponds to a Cuda device.");
 	helpLines.push_back("\t\tThe memory structure contains the total memory on the device and the memory available for a Cuda call.");
 }
\ No newline at end of file

src/c/Mex/MexDeviceStats.cpp

+#include "MexCommand.h"
+#include "CWrappers.h"
+#include "../Cuda/CudaDeviceStats.h"
+
+void MexDeviceStats::execute(int nlhs, mxArray* plhs[], int nrhs, const mxArray* prhs[]) const
+{
+    DevStats* devStats;
+    int numDevices = deviceStats(&devStats);
+
+    mwSize dims[2] = {numDevices,1};
+    const char* fieldNames[] = {"name", "major", "minor", "constMem", "sharedMem", "totalMem", "tccDriver", "mpCount", "threadsPerMP", "warpSize", "maxThreads"};
+    plhs[0] = mxCreateStructArray(2, dims, 11, fieldNames);
+    int name_field = mxGetFieldNumber(plhs[0], "name");
+    int major_field = mxGetFieldNumber(plhs[0], "major");
+    int minor_field = mxGetFieldNumber(plhs[0], "minor");
+    int constMem_field = mxGetFieldNumber(plhs[0], "constMem");
+    int sharedMem_field = mxGetFieldNumber(plhs[0], "sharedMem");
+    int totalMem_field = mxGetFieldNumber(plhs[0], "totalMem");
+    int tccDriver_field = mxGetFieldNumber(plhs[0], "tccDriver");
+    int mpCount_field = mxGetFieldNumber(plhs[0], "mpCount");
+    int threadsPerMP_field = mxGetFieldNumber(plhs[0], "threadsPerMP");
+    int warpSize_field = mxGetFieldNumber(plhs[0], "warpSize");
+    int maxThreads_field = mxGetFieldNumber(plhs[0], "maxThreads");
+
+    for(int device = 0; device<numDevices; ++device)
+    {
+        DevStats curDevice = devStats[device];
+
+        mxSetFieldByNumber(plhs[0], device, name_field, mxCreateString(curDevice.name.c_str()));
+        mxSetFieldByNumber(plhs[0], device, major_field, mxCreateDoubleScalar(curDevice.major));
+        mxSetFieldByNumber(plhs[0], device, minor_field, mxCreateDoubleScalar(curDevice.minor));
+        mxSetFieldByNumber(plhs[0], device, constMem_field, mxCreateDoubleScalar(curDevice.constMem));
+        mxSetFieldByNumber(plhs[0], device, sharedMem_field, mxCreateDoubleScalar(curDevice.sharedMem));
+        mxSetFieldByNumber(plhs[0], device, totalMem_field, mxCreateDoubleScalar(curDevice.totalMem));
+        mxSetFieldByNumber(plhs[0], device, tccDriver_field, mxCreateLogicalScalar(curDevice.tccDriver));
+        mxSetFieldByNumber(plhs[0], device, mpCount_field, mxCreateDoubleScalar(curDevice.mpCount));
+        mxSetFieldByNumber(plhs[0], device, threadsPerMP_field, mxCreateDoubleScalar(curDevice.threadsPerMP));
+        mxSetFieldByNumber(plhs[0], device, warpSize_field, mxCreateDoubleScalar(curDevice.warpSize));
+        mxSetFieldByNumber(plhs[0], device, maxThreads_field, mxCreateDoubleScalar(curDevice.maxThreads));
+    }
+
+    delete[] devStats;
+}
+
+std::string MexDeviceStats::check(int nlhs, mxArray* plhs[], int nrhs, const mxArray* prhs[]) const
+{
+    if(nlhs<1)
+        return "Requires one output!";
+
+    return "";
+}
+
+void MexDeviceStats::usage(std::vector<std::string>& outArgs, std::vector<std::string>& inArgs) const
+{
+    outArgs.push_back("deviceStatsArray");
+}
+
+void MexDeviceStats::help(std::vector<std::string>& helpLines) const
+{
+    helpLines.push_back("This will return the statistics of each Cuda capable device installed.");
+
+    helpLines.push_back("\tDeviceStatsArray -- this is an array of structs, one struct per device.");
+    helpLines.push_back("\t\tThe struct has these fields: name, major, minor, constMem, sharedMem, totalMem, tccDriver, mpCount, threadsPerMP, warpSize, maxThreads.");
+}