CUDA7.5 support for older cards

nicehash · Nov 3, 2016 · 32d6b6a · 32d6b6a
1 parent e769c27
commit 32d6b6a
Show file tree

Hide file tree

Showing 11 changed files with 198 additions and 55 deletions.
diff --git a/cuda_tromp/cuda_tromp.cpp b/cuda_tromp/cuda_tromp.cpp
@@ -10,7 +10,7 @@ struct proof;
 #include "eqcuda.hpp"
 
 
-cuda_tromp::cuda_tromp(int platf_id, int dev_id)
+SOLVER_NAME::SOLVER_NAME(int platf_id, int dev_id)
 {
 	device_id = dev_id;
 	getinfo(0, dev_id, m_gpu_name, m_sm_count, m_version);
@@ -21,21 +21,21 @@ cuda_tromp::cuda_tromp(int platf_id, int dev_id)
 }
 
 
-std::string cuda_tromp::getdevinfo()
+std::string SOLVER_NAME::getdevinfo()
 {
 	return m_gpu_name + " (#" + std::to_string(device_id) + ") BLOCKS=" + 
 		std::to_string(blocks) + ", THREADS=" + std::to_string(threadsperblock);
 }
 
 
-int cuda_tromp::getcount()
+int SOLVER_NAME::getcount()
 {
 	int device_count;
 	checkCudaErrors(cudaGetDeviceCount(&device_count));
 	return device_count;
 }
 
-void cuda_tromp::getinfo(int platf_id, int d_id, std::string& gpu_name, int& sm_count, std::string& version)
+void SOLVER_NAME::getinfo(int platf_id, int d_id, std::string& gpu_name, int& sm_count, std::string& version)
 {
 	//int runtime_version;
 	//checkCudaErrors(cudaRuntimeGetVersion(&runtime_version));
@@ -50,27 +50,27 @@ void cuda_tromp::getinfo(int platf_id, int d_id, std::string& gpu_name, int& sm_
 }
 
 
-void cuda_tromp::start(cuda_tromp& device_context) 
+void SOLVER_NAME::start(SOLVER_NAME& device_context)
 { 
 	device_context.context = new eq_cuda_context(device_context.threadsperblock, 
 		device_context.blocks,
 		device_context.device_id);
 }
 
-void cuda_tromp::stop(cuda_tromp& device_context) 
+void SOLVER_NAME::stop(SOLVER_NAME& device_context)
 { 
 	if (device_context.context)
 		delete device_context.context;
 }
 
-void cuda_tromp::solve(const char *tequihash_header,
+void SOLVER_NAME::solve(const char *tequihash_header,
 	unsigned int tequihash_header_len,
 	const char* nonce,
 	unsigned int nonce_len,
 	std::function<bool()> cancelf,
 	std::function<void(const std::vector<uint32_t>&, size_t, const unsigned char*)> solutionf,
 	std::function<void(void)> hashdonef,
-	cuda_tromp& device_context)
+	SOLVER_NAME& device_context)
 {
 	device_context.context->solve(tequihash_header,
 		tequihash_header_len,

diff --git a/cuda_tromp/cuda_tromp.hpp b/cuda_tromp/cuda_tromp.hpp
@@ -4,26 +4,32 @@
 #define DLL_CUDA_TROMP
 #endif
 
+#ifdef CUDA_75
+#define SOLVER_NAME cuda_tromp_75
+#else
+#define SOLVER_NAME cuda_tromp
+#endif
+
 struct eq_cuda_context;
 
-struct DLL_CUDA_TROMP cuda_tromp
+struct DLL_CUDA_TROMP SOLVER_NAME
 {
 	int threadsperblock;
 	int blocks;
 	int device_id;
 	eq_cuda_context* context;
 
-	cuda_tromp(int platf_id, int dev_id);
+	SOLVER_NAME(int platf_id, int dev_id);
 
 	std::string getdevinfo();
 
 	static int getcount();
 
 	static void getinfo(int platf_id, int d_id, std::string& gpu_name, int& sm_count, std::string& version);
 
-	static void start(cuda_tromp& device_context);
+	static void start(SOLVER_NAME& device_context);
 
-	static void stop(cuda_tromp& device_context);
+	static void stop(SOLVER_NAME& device_context);
 
 	static void solve(const char *tequihash_header,
 		unsigned int tequihash_header_len,
@@ -32,12 +38,54 @@ struct DLL_CUDA_TROMP cuda_tromp
 		std::function<bool()> cancelf,
 		std::function<void(const std::vector<uint32_t>&, size_t, const unsigned char*)> solutionf,
 		std::function<void(void)> hashdonef,
-		cuda_tromp& device_context);
+		SOLVER_NAME& device_context);
 
 	std::string getname() { return "CUDA-TROMP"; }
 
 private:
 	std::string m_gpu_name;
 	std::string m_version;
 	int m_sm_count;
-};
+};
+
+#ifndef _LIB
+#undef SOLVER_NAME
+#define SOLVER_NAME cuda_tromp_75
+
+struct DLL_CUDA_TROMP SOLVER_NAME
+{
+	int threadsperblock;
+	int blocks;
+	int device_id;
+	eq_cuda_context* context;
+
+	SOLVER_NAME(int platf_id, int dev_id);
+
+	std::string getdevinfo();
+
+	static int getcount();
+
+	static void getinfo(int platf_id, int d_id, std::string& gpu_name, int& sm_count, std::string& version);
+
+	static void start(SOLVER_NAME& device_context);
+
+	static void stop(SOLVER_NAME& device_context);
+
+	static void solve(const char *tequihash_header,
+		unsigned int tequihash_header_len,
+		const char* nonce,
+		unsigned int nonce_len,
+		std::function<bool()> cancelf,
+		std::function<void(const std::vector<uint32_t>&, size_t, const unsigned char*)> solutionf,
+		std::function<void(void)> hashdonef,
+		SOLVER_NAME& device_context);
+
+	std::string getname() { return "CUDA-TROMP-75"; }
+
+private:
+	std::string m_gpu_name;
+	std::string m_version;
+	int m_sm_count;
+};
+
+#endif
diff --git a/cuda_tromp/cuda_tromp.vcxproj b/cuda_tromp/cuda_tromp.vcxproj
@@ -5,6 +5,10 @@
       <Configuration>Debug</Configuration>
       <Platform>x64</Platform>
     </ProjectConfiguration>
+    <ProjectConfiguration Include="Release7.5|x64">
+      <Configuration>Release7.5</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
     <ProjectConfiguration Include="Release|x64">
       <Configuration>Release</Configuration>
       <Platform>x64</Platform>
@@ -26,6 +30,7 @@
   <PropertyGroup Label="Globals">
     <ProjectGuid>{33C2B469-F025-4223-B9B6-E69D42FEA7D6}</ProjectGuid>
     <RootNamespace>cuda_tromp</RootNamespace>
+    <CudaToolkitCustomDir>$(CUDA_PATH_V7_5)</CudaToolkitCustomDir>
   </PropertyGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
@@ -41,6 +46,13 @@
     <CharacterSet>MultiByte</CharacterSet>
     <PlatformToolset>v120</PlatformToolset>
   </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release7.5|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+    <PlatformToolset>v120</PlatformToolset>
+  </PropertyGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
   <ImportGroup Label="ExtensionSettings">
     <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 8.0.props" />
@@ -51,10 +63,18 @@
   <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
     <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
   </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release7.5|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
   <PropertyGroup Label="UserMacros" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
     <LinkIncremental>true</LinkIncremental>
   </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release7.5|x64'">
+    <TargetName>$(ProjectName)_75</TargetName>
+    <OutDir>$(SolutionDir)$(Platform)\Release\</OutDir>
+    <IntDir>$(Platform)\Release\</IntDir>
+  </PropertyGroup>
   <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
     <ClCompile>
       <WarningLevel>Level3</WarningLevel>
@@ -105,7 +125,35 @@ copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
     </PostBuildEvent>
     <CudaCompile>
       <TargetMachinePlatform>64</TargetMachinePlatform>
-      <CodeGeneration>compute_50,sm_50</CodeGeneration>
+      <CodeGeneration>compute_61,sm_61;compute_52,sm_52;compute_50,sm_50;compute_35,sm_35;compute_30,sm_30;</CodeGeneration>
+      <PtxAsOptionV>true</PtxAsOptionV>
+    </CudaCompile>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release7.5|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;WIN64;NDEBUG;_CONSOLE;_LIB;CUDA_75;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <MultiProcessorCompilation>true</MultiProcessorCompilation>
+      <DisableSpecificWarnings>4334;4316;4244;4996;4251;</DisableSpecificWarnings>
+      <AdditionalIncludeDirectories>..\3rdparty\include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <SubSystem>Console</SubSystem>
+      <AdditionalDependencies>cudart.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+    <PostBuildEvent>
+      <Command>echo copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"
+copy "$(CudaToolkitBinDir)\cudart*.dll" "$(OutDir)"</Command>
+    </PostBuildEvent>
+    <CudaCompile>
+      <TargetMachinePlatform>64</TargetMachinePlatform>
+      <CodeGeneration>compute_52,sm_52;compute_50,sm_50;compute_35,sm_35;compute_30,sm_30;compute_20,sm_21;</CodeGeneration>
       <PtxAsOptionV>true</PtxAsOptionV>
     </CudaCompile>
   </ItemDefinitionGroup>

diff --git a/nheqminer/libstratum/StratumClient.cpp b/nheqminer/libstratum/StratumClient.cpp
@@ -421,6 +421,8 @@ bool StratumClient<Miner, Job, Solution>::submit(const Solution* solution, const
 	return true;
 }
 
-template class StratumClient<ZMinerAVX, ZcashJob, EquihashSolution>;
-template class StratumClient<ZMinerSSE2, ZcashJob, EquihashSolution>;
+template class StratumClient<ZMinerAVXCUDA80, ZcashJob, EquihashSolution>;
+template class StratumClient<ZMinerSSE2CUDA80, ZcashJob, EquihashSolution>;
+template class StratumClient<ZMinerAVXCUDA75, ZcashJob, EquihashSolution>;
+template class StratumClient<ZMinerSSE2CUDA75, ZcashJob, EquihashSolution>;
 
diff --git a/nheqminer/libstratum/StratumClient.h b/nheqminer/libstratum/StratumClient.h
@@ -162,6 +162,8 @@ class StratumClient
 };
 
 
-typedef StratumClient<ZMinerAVX, ZcashJob, EquihashSolution> ZcashStratumClientAVX;
-typedef StratumClient<ZMinerSSE2, ZcashJob, EquihashSolution> ZcashStratumClientSSE2;
+typedef StratumClient<ZMinerAVXCUDA80, ZcashJob, EquihashSolution> ZcashStratumClientAVXCUDA80;
+typedef StratumClient<ZMinerSSE2CUDA80, ZcashJob, EquihashSolution> ZcashStratumClientSSE2CUDA80;
+typedef StratumClient<ZMinerAVXCUDA75, ZcashJob, EquihashSolution> ZcashStratumClientAVXCUDA75;
+typedef StratumClient<ZMinerSSE2CUDA75, ZcashJob, EquihashSolution> ZcashStratumClientSSE2CUDA75;
 
diff --git a/nheqminer/libstratum/ZcashStratum.cpp b/nheqminer/libstratum/ZcashStratum.cpp
@@ -603,6 +603,8 @@ void ZcashMiner<CPUSolver, CUDASolver, OPENCLSolver>::failedSolution()
 
 template class ZcashMiner<cpu_xenoncat, cuda_tromp, ocl_xmp>;
 template class ZcashMiner<cpu_tromp, cuda_tromp, ocl_xmp>;
+template class ZcashMiner<cpu_xenoncat, cuda_tromp_75, ocl_xmp>;
+template class ZcashMiner<cpu_tromp, cuda_tromp_75, ocl_xmp>;
 
 std::mutex benchmark_work;
 std::vector<uint256*> benchmark_nonces;
@@ -808,12 +810,12 @@ void ZcashMiner<CPUSolver, CUDASolver, OPENCLSolver>::doBenchmark(int hashes, in
 }
 
 
-void ZMinerAVX_doBenchmark(int hashes, int cpu_threads, int cuda_count, int* cuda_en, int* cuda_b, int* cuda_t,
-                           int opencl_count, int opencl_platf, int* opencl_en) {
-    ZMinerAVX::doBenchmark(hashes, cpu_threads, cuda_count, cuda_en, cuda_b, cuda_t, opencl_count, opencl_platf, opencl_en);
-}
-
-void ZMinerSSE2_doBenchmark(int hashes, int cpu_threads, int cuda_count, int* cuda_en, int* cuda_b, int* cuda_t,
-                            int opencl_count, int opencl_platf, int* opencl_en) {
-    ZMinerSSE2::doBenchmark(hashes, cpu_threads, cuda_count, cuda_en, cuda_b, cuda_t, opencl_count, opencl_platf, opencl_en);
-}
+//void ZMinerAVX_doBenchmark(int hashes, int cpu_threads, int cuda_count, int* cuda_en, int* cuda_b, int* cuda_t,
+//                           int opencl_count, int opencl_platf, int* opencl_en) {
+//    ZMinerAVX::doBenchmark(hashes, cpu_threads, cuda_count, cuda_en, cuda_b, cuda_t, opencl_count, opencl_platf, opencl_en);
+//}
+//
+//void ZMinerSSE2_doBenchmark(int hashes, int cpu_threads, int cuda_count, int* cuda_en, int* cuda_b, int* cuda_t,
+//                            int opencl_count, int opencl_platf, int* opencl_en) {
+//    ZMinerSSE2::doBenchmark(hashes, cpu_threads, cuda_count, cuda_en, cuda_b, cuda_t, opencl_count, opencl_platf, opencl_en);
+//}
diff --git a/nheqminer/libstratum/ZcashStratum.h b/nheqminer/libstratum/ZcashStratum.h
@@ -141,8 +141,10 @@ class ZcashMiner
 		int opencl_count, int opencl_platf, int* opencl_en);
 };
 
-typedef ZcashMiner<cpu_xenoncat, cuda_tromp, ocl_xmp> ZMinerAVX;
-typedef ZcashMiner<cpu_tromp, cuda_tromp, ocl_xmp> ZMinerSSE2;
+typedef ZcashMiner<cpu_xenoncat, cuda_tromp, ocl_xmp> ZMinerAVXCUDA80;
+typedef ZcashMiner<cpu_tromp, cuda_tromp, ocl_xmp> ZMinerSSE2CUDA80;
+typedef ZcashMiner<cpu_xenoncat, cuda_tromp_75, ocl_xmp> ZMinerAVXCUDA75;
+typedef ZcashMiner<cpu_tromp, cuda_tromp_75, ocl_xmp> ZMinerSSE2CUDA75;
 
 // gcc static undefined reference workaround
 void ZMinerAVX_doBenchmark(int hashes, int cpu_threads, int cuda_count, int* cuda_en, int* cuda_b, int* cuda_t,