added compilation for compute 5.2,6.0 and 6.1

prime9 · Jul 13, 2017 · bfc662d · bfc662d
1 parent bd48812
commit bfc662d
Show file tree

Hide file tree

Showing 7 changed files with 18 additions and 7 deletions.
diff --git a/.gitignore b/.gitignore
@@ -8,6 +8,7 @@
 *.orig
 \#*
 .\#*
+.vscode
 
 # Local build configuration
 Config.make

diff --git a/CNTK.Cpp.props b/CNTK.Cpp.props
@@ -14,6 +14,12 @@
     <NvmlLibPath Condition="'$(CudaVersion)' == '7.5'">"c:\Program Files\NVIDIA Corporation\GDK\gdk_win7_amd64_release\nvml\lib"</NvmlLibPath>
     <NvmlLibPath Condition="'$(CudaVersion)' == '8.0'" />
 
+    <NvidiaCompute />
+    <!-- see http://docs.nvidia.com/cuda/pascal-compatibility-guide/index.html#building-applications-with-pascal-support -->
+    <NvidiaCompute Condition="$(DebugBuild)">compute_30,sm_30</NvidiaCompute>
+    <NvidiaCompute Condition="$(ReleaseBuild) And '$(CudaVersion)' == '7.5'">compute_30,sm_30;compute_35,sm_35;compute_50,sm_50;compute_52,sm_52;</NvidiaCompute>
+    <NvidiaCompute Condition="$(ReleaseBuild) And '$(CudaVersion)' == '8.0'">compute_30,sm_30;compute_35,sm_35;compute_50,sm_50;compute_52,sm_52;compute_61,sm_61;</NvidiaCompute>
+
     <NvmlDll>%ProgramW6432%\NVIDIA Corporation\NVSMI\nvml.dll</NvmlDll>
     <NvmlDll Condition="Exists('c:\local\bindrop\NVSMI\nvml.dll')">c:\local\bindrop\NVSMI\nvml.dll</NvmlDll>
 

diff --git a/Makefile b/Makefile
@@ -214,9 +214,13 @@ endif
 
 # Set up nvcc target architectures (will generate code to support them all, i.e. fat-binary, in release mode)
 # In debug mode we will rely on JIT to create code "on the fly" for the underlying architecture
+# see also http://docs.nvidia.com/cuda/pascal-compatibility-guide/index.html#building-applications-with-pascal-support
 GENCODE_SM30 := -gencode arch=compute_30,code=\"sm_30,compute_30\"
 GENCODE_SM35 := -gencode arch=compute_35,code=\"sm_35,compute_35\"
 GENCODE_SM50 := -gencode arch=compute_50,code=\"sm_50,compute_50\"
+GENCODE_SM52 := -gencode arch=compute_52,code=\"sm_52,compute_52\"
+GENCODE_SM60 := -gencode arch=compute_60,code=\"sm_60,compute_60\"
+GENCODE_SM61 := -gencode arch=compute_61,code=\"sm_61,compute_61\"
 
 # Should we relocate *.gcno and *.gcda files using -fprofile-dir option?
 # Use GCOV_PREFIX and GCOV_PREFIX_STRIP if relocating:
@@ -246,7 +250,7 @@ ifeq ("$(BUILDTYPE)","release")
   ifdef CNTK_CUDA_CODEGEN_RELEASE
     GENCODE_FLAGS := $(CNTK_CUDA_CODEGEN_RELEASE)
   else
-    GENCODE_FLAGS := $(GENCODE_SM30) $(GENCODE_SM35) $(GENCODE_SM50)
+    GENCODE_FLAGS := $(GENCODE_SM30) $(GENCODE_SM35) $(GENCODE_SM50) $(GENCODE_SM52) $(GENCODE_SM60) $(GENCODE_SM61)
   endif
 
   CXXFLAGS += -g -O4

diff --git a/Source/Math/Math.vcxproj b/Source/Math/Math.vcxproj
@@ -86,7 +86,7 @@
     </PostBuildEvent>
     <CudaCompile>
       <TargetMachinePlatform>64</TargetMachinePlatform>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;compute_50,sm_50;</CodeGeneration>
+      <CodeGeneration>$(NvidiaCompute);</CodeGeneration>
       <CudaRuntime>None</CudaRuntime>
       <NvccCompilation>compile</NvccCompilation>
       <GenerateLineInfo>true</GenerateLineInfo>
@@ -124,7 +124,7 @@
     <CudaCompile>
       <FastMath>true</FastMath>
       <TargetMachinePlatform>64</TargetMachinePlatform>
-      <CodeGeneration>compute_30,sm_30;compute_35,sm_35</CodeGeneration>
+      <CodeGeneration>compute_30,sm_30;compute_35,sm_35;$(NvidiaCompute)</CodeGeneration>
       <GPUDebugInfo>true</GPUDebugInfo>
       <GenerateLineInfo>true</GenerateLineInfo>
       <HostDebugInfo>true</HostDebugInfo>

diff --git a/Source/Math/MathCUDA.vcxproj b/Source/Math/MathCUDA.vcxproj
@@ -56,11 +56,11 @@
   <PropertyGroup Label="UserMacros" />
   <PropertyGroup Condition="$(DebugBuild)">
     <CudaCodeGen>$(CNTK_CUDA_CODEGEN_DEBUG)</CudaCodeGen>
-    <CudaCodeGen Condition="'$(CudaCodeGen)'==''">compute_30,sm_30</CudaCodeGen>
+    <CudaCodeGen Condition="'$(CudaCodeGen)'==''">$(NvidiaCompute)</CudaCodeGen>
   </PropertyGroup>
   <PropertyGroup Condition="$(ReleaseBuild)">
     <CudaCodeGen>$(CNTK_CUDA_CODEGEN_RELEASE)</CudaCodeGen>
-    <CudaCodeGen Condition="'$(CudaCodeGen)'==''">compute_30,sm_30;compute_35,sm_35;compute_50,sm_50;</CudaCodeGen>
+    <CudaCodeGen Condition="'$(CudaCodeGen)'==''">$(NvidiaCompute)</CudaCodeGen>
   </PropertyGroup>
   <PropertyGroup>
     <TargetName>Cntk.Math.Cuda-$(CntkComponentVersion)</TargetName>

diff --git a/Tests/UnitTests/EvalTests/EvalTests.vcxproj b/Tests/UnitTests/EvalTests/EvalTests.vcxproj
@@ -82,7 +82,7 @@
     </Link>
     <CudaCompile>
       <TargetMachinePlatform>64</TargetMachinePlatform>
-      <CodeGeneration>compute_30,sm_30;%(CodeGeneration)</CodeGeneration>
+      <CodeGeneration>$(NvidiaCompute)</CodeGeneration>
     </CudaCompile>
   </ItemDefinitionGroup>
   <ItemDefinitionGroup Condition="$(ReleaseBuild)">

diff --git a/Tests/UnitTests/MathPerformanceTests/MathPerformanceTests.vcxproj b/Tests/UnitTests/MathPerformanceTests/MathPerformanceTests.vcxproj
@@ -74,7 +74,7 @@
     </Link>
     <CudaCompile>
       <TargetMachinePlatform>64</TargetMachinePlatform>
-      <CodeGeneration>compute_30,sm_30;%(CodeGeneration)</CodeGeneration>
+      <CodeGeneration>%(CodeGeneration)</CodeGeneration>
     </CudaCompile>
   </ItemDefinitionGroup>
   <ItemDefinitionGroup Condition="$(ReleaseBuild)">
-Original file line number
+Diff line change
@@ Expand Up / @@ -8,6 +8,7 @@ @@
     *.orig
     \#*
     .\#*
+    .vscode
     # Local build configuration
     Config.make
@@ Expand Down @@