Fix Mac compatibility

Changed unix and mac checking, as unix=linux&mac Aslo added xml.
CERN · Oct 5, 2016 · 87110b9 · 87110b9
1 parent c217793
commit 87110b9
Show file tree

Hide file tree

Showing 5 changed files with 136 additions and 30 deletions.
diff --git a/Compile.m b/Compile.m
@@ -44,21 +44,10 @@
         mex  ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/win32
         mex  ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/win32
     end
-elseif isunix
-    if ~isempty(strfind(computer('arch'),'64'))
-        mex -largeArrayDims ./Source/Ax.cpp ./Source/ray_interpolated_projection.cu ./Source/Siddon_projection.cu ./Source/ray_interpolated_projection_parallel.cu ./Source/Siddon_projection_parallel.cu -outdir ./Mex_files/linux64
-        mex -largeArrayDims ./Source/Atb.cpp ./Source/voxel_backprojection.cu ./Source/voxel_backprojection2.cu ./Source/voxel_backprojection_parallel.cu -outdir ./Mex_files/linux64
-        mex -largeArrayDims ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/linux64
-        mex -largeArrayDims ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/linux64
-   else
-        mex  ./Source/Ax.cpp ./Source/ray_interpolated_projection.cu ./Source/Siddon_projection.cu ./Source/ray_interpolated_projection_parallel.cu ./Source/Siddon_projection_parallel.cu -outdir ./Mex_files/linux32
-        mex  ./Source/Atb.cpp ./Source/voxel_backprojection.cu ./Source/voxel_backprojection2.cu ./Source/voxel_backprojection_parallel.cu -outdir ./Mex_files/linux32
-        mex  ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/linux32
-        mex  ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/linux32
-    end
 
 elseif ismac
     if ~isempty(strfind(computer('arch'),'64'))
+        disp('compiling for mac 64')
         mex -largeArrayDims ./Source/Ax.cpp ./Source/ray_interpolated_projection.cu ./Source/Siddon_projection.cu ./Source/ray_interpolated_projection_parallel.cu ./Source/Siddon_projection_parallel.cu -outdir ./Mex_files/mac64
         mex -largeArrayDims ./Source/Atb.cpp ./Source/voxel_backprojection.cu ./Source/voxel_backprojection2.cu ./Source/voxel_backprojection_parallel.cu -outdir ./Mex_files/mac64
         mex -largeArrayDims ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/mac64
@@ -69,11 +58,24 @@
         mex  ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/mac32
         mex  ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/mac32
     end
+
+elseif isunix
+    if ~isempty(strfind(computer('arch'),'64'))
+        mex -largeArrayDims ./Source/Ax.cpp ./Source/ray_interpolated_projection.cu ./Source/Siddon_projection.cu ./Source/ray_interpolated_projection_parallel.cu ./Source/Siddon_projection_parallel.cu -outdir ./Mex_files/linux64
+        mex -largeArrayDims ./Source/Atb.cpp ./Source/voxel_backprojection.cu ./Source/voxel_backprojection2.cu ./Source/voxel_backprojection_parallel.cu -outdir ./Mex_files/linux64
+        mex -largeArrayDims ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/linux64
+        mex -largeArrayDims ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/linux64
+   else
+        mex  ./Source/Ax.cpp ./Source/ray_interpolated_projection.cu ./Source/Siddon_projection.cu ./Source/ray_interpolated_projection_parallel.cu ./Source/Siddon_projection_parallel.cu -outdir ./Mex_files/linux32
+        mex  ./Source/Atb.cpp ./Source/voxel_backprojection.cu ./Source/voxel_backprojection2.cu ./Source/voxel_backprojection_parallel.cu -outdir ./Mex_files/linux32
+        mex  ./Source/minTV.cpp ./Source/POCS_TV.cu  -outdir ./Mex_files/linux32
+        mex  ./Source/tvDenoise.cpp ./Source/tvdenoising.cu  -outdir ./Mex_files/linux32
+   end
 end
 
 
 
 
 
 disp('')
-disp('Compilation complete')
+disp('Compilation complete')
diff --git a/InitTIGRE.m b/InitTIGRE.m
@@ -25,12 +25,26 @@
 addpath(genpath('./Test_data'));
 
 % different arch versions
-if ~isempty(strfind(computer('arch'),'64'))
-    addpath('./Mex_files/win64');
+if ispc
+    if ~isempty(strfind(computer('arch'),'64'))
+        addpath('./Mex_files/win64');
+    else
+        addpath('./Mex_files/win32');
+    end
+elseif ismac
+    if ~isempty(strfind(computer('arch'),'64'))
+        addpath('./Mex_files/mac64');
+    else
+        addpath('./Mex_files/mac32');
+    end
 else
-    addpath('./Mex_files/win32');
+    if ~isempty(strfind(computer('arch'),'64'))
+        addpath('./Mex_files/linux64');
+    else
+        addpath('./Mex_files/linux32');
+    end
 end
-
+    
 addpath('./Demos');
 
 % Perceptually uniform colormaps
@@ -40,25 +54,24 @@
 addpath('./Third_party_tools/sec2hours');
 addpath('./Third_party_tools/readMHD');
 
-
 if ispc
     if ~isempty(strfind(computer('arch'),'64'))
         addpath('./Mex_files/win64');
     else
         addpath('./Mex_files/win32');
     end
-elseif isunix
-    if ~isempty(strfind(computer('arch'),'64'))
-        addpath('./Mex_files/linux64');
-    else
-        addpath('./Mex_files/linux32');
-    end
 elseif ismac
     if ~isempty(strfind(computer('arch'),'64'))
         addpath('./Mex_files/mac64');
     else
         addpath('./Mex_files/mac32');
     end
+else
+    if ~isempty(strfind(computer('arch'),'64'))
+        addpath('./Mex_files/linux64');
+    else
+        addpath('./Mex_files/linux32');
+    end
 end
 
 if ispc
@@ -75,4 +88,4 @@
     warning('TIGRE needs a big amount of memory, be careful when running big images.')
 end
 
-clear all;
+clear all;
diff --git a/Source/POCS_TV.cu b/Source/POCS_TV.cu
@@ -325,7 +325,7 @@ do { \
         cudaMemcpy(dst, d_image, mem_size, cudaMemcpyDeviceToHost);
         cudaCheckErrors("Copy result back");
 
-        cudaFree(b);
+        cudaFree(d_image);
         cudaFree(d_norm2aux);
         cudaFree(d_dimgTV);
         cudaFree(d_norm2);

diff --git a/Source/Siddon_projection_parallel.cu b/Source/Siddon_projection_parallel.cu
@@ -170,7 +170,7 @@ __global__ void kernelPixelDetector_parallel( Geometry geo,
 
     // Compute max/min image INDEX for intersection eq(11-19)
     // Discussion about ternary operator in CUDA: https://stackoverflow.com/questions/7104384/in-cuda-why-is-a-b010-more-efficient-than-an-if-else-version
-    float imin,imax,jmin,jmax,kmin,kmax;
+    float imin,imax,jmin,jmax;
     // for X
     if( source.x<pixel1D.x){
         imin=(am==axm)? 1           : ceil (source.x+am*ray.x);
@@ -197,7 +197,7 @@ __global__ void kernelPixelDetector_parallel( Geometry geo,
 //     }
 
     // get intersection point N1. eq(20-21) [(also eq 9-10)]
-    float ax,ay,az;
+    float ax,ay;
     ax=(source.x<pixel1D.x)?  (imin-source.x)/ray.x  :  (imax-source.x)/ray.x;
     ay=(source.y<pixel1D.y)?  (jmin-source.y)/ray.y  :  (jmax-source.y)/ray.y;
 //     az=(source.z<pixel1D.z)?  (kmin-source.z)/ray.z  :  (kmax-source.z)/ray.z;
@@ -214,12 +214,12 @@ __global__ void kernelPixelDetector_parallel( Geometry geo,
     // Initialize
     float ac=am;
     //eq (28), unit alphas
-    float axu,ayu,azu;
+    float axu,ayu;
     axu=1/abs(ray.x);
     ayu=1/abs(ray.y);
 //     azu=1/abs(ray.z);
     // eq(29), direction of update
-    float iu,ju,ku;
+    float iu,ju;
     iu=(source.x< pixel1D.x)? 1 : -1;
     ju=(source.y< pixel1D.y)? 1 : -1;
 //     ku=(source.z< pixel1D.z)? 1 : -1;

diff --git a/mex_CUDA_maci64.xml b/mex_CUDA_maci64.xml
@@ -0,0 +1,91 @@
+<?xml version="1.0" encoding="UTF-8" ?>
+<!-- Copyright 2013 The MathWorks, Inc. -->
+<config
+    Name="nvcc"
+    ShortName="nvcc"
+    Manufacturer="NVIDIA"
+    Version=""
+    Language="CUDA"
+    Priority="A"
+    Location="$NVCC" >
+    <Details
+        CompilerExecutable="$CXX"
+        CompilerDefines="$DEFINES"
+        CompilerFlags="$CXXFLAGS"
+        OptimizationFlags="$CXXOPTIMFLAGS"
+        DebugFlags="$CXXDEBUGFLAGS"
+        IncludeFlags="$INCLUDE"
+        LinkerLibraries="$LINKLIBS"
+        LinkerOptimizationFlags="$LDOPTIMFLAGS"
+        LinkerDebugFlags="$LDDEBUGFLAGS"
+
+        CompilerDefineFormatter="-D%s"
+        LinkerLibrarySwitchFormatter="-l%s"
+        LinkerPathFormatter="-L%s"
+    />
+    <vars
+          CMDLINE1="$CXX -c $DEFINES $INCLUDE $NVCCFLAGS $CXXFLAGS $OPTIM $SRC -o $OBJ"
+          CMDLINE2="$LDXX $LDFLAGS $LINKOPTIM $LINKEXPORT $OBJS $CXXLIBS $LINKLIBS -o $EXE"
+          CMDLINE3="rm -f $OBJ"
+
+          CXX="$NVCC"
+          DEFINES="$MATLABMEX"
+          MATLABMEX="-DMATLAB_MEX_FILE"
+          NVCCFLAGS="-gencode=arch=compute_20,code=sm_20 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_35,code=sm_35 -gencode=arch=compute_50,code=sm_50 -m 64 $NVCC_FLAGS"
+          CXXFLAGS="--compiler-bindir=$CLANG_COMPILER --compiler-options=-fno-common,-arch,$ARCHS,-mmacosx-version-min=$MACOSX_DEPLOYMENT_TARGET,-fexceptions,-fPIC,-fno-omit-frame-pointer,-pthread,-isysroot,$ISYSROOT"
+          INCLUDE="-I&quot;$MATLABROOT/extern/include&quot; -I&quot;$MATLABROOT/simulink/include&quot; -I&quot;$MATLABROOT/toolbox/distcomp/gpu/extern/include/&quot;" 
+          CXXOPTIMFLAGS="-O2 -DNDEBUG"
+          CXXDEBUGFLAGS="-g"
+          ARCHS="x86_64"
+
+          LDXX="$XCRUN_DIR/xcrun clang++"
+          LDFLAGS="-stdlib=libstdc++ -pthread -Wl,-twolevel_namespace -undefined error -arch $ARCHS -mmacosx-version-min=$MACOSX_DEPLOYMENT_TARGET -Wl,-syslibroot,$ISYSROOT $LDBUNDLE $LINKEXPORT"
+          LDBUNDLE="-bundle"
+          LINKEXPORT="-Wl,-exported_symbols_list,&quot;$MATLABROOT/extern/lib/$ARCH/mexFunction.map&quot;"
+          LINKLIBS="-L&quot;$MATLABROOT/bin/$ARCH&quot; -lmx -lmex -lmat -lm -lmwgpu -lcudart"
+          LDOPTIMFLAGS="-O"
+          LDDEBUGFLAGS="-g"
+
+          OBJEXT=".o"
+          LDEXT=".mexmaci64"
+          MACOSX_DEPLOYMENT_TARGET='10.11'>
+    </vars>
+    <client>
+    </client>
+    <locationFinder>
+        <NVCC>
+            <or>
+                <and>
+                    <envVarExists name="MW_NVCC_PATH" />
+                    <dirExists name="$$" />
+                    <fileExists name="$$/nvcc" />
+                </and>
+                <cmdReturns name="which nvcc" />
+                <fileExists name="/Developer/NVIDIA/CUDA-7.5/bin/nvcc" /> <!-- Default install location -->
+                <fileExists name="/usr/local/cuda/bin/nvcc" />            <!-- Recent CUDA versions symlink here -->
+            </or>
+        </NVCC>
+		<CLANG_COMPILER>
+            <or>
+                <envVarExists name="MW_XCODE_CLANG_COMPILER" />
+                <and>
+                    <cmdReturns name="xcrun -f clang++" />
+                    <dirExists name="$$" />
+                    <fileExists name="$$/clang++" />
+                </and>
+            </or>
+        </CLANG_COMPILER>
+        <XCRUN_DIR>
+            <and>
+                <cmdReturns name="which xcrun" />
+                <dirExists name="$$" />
+            </and>
+        </XCRUN_DIR>
+        <ISYSROOT>
+          <and>
+              <cmdReturns name="xcrun --show-sdk-path"/>
+              <dirExists name="$$" />
+         </and>
+        </ISYSROOT>
+    </locationFinder>
+</config>