mirror of
https://github.com/NVIDIA/cuda-samples.git
synced 2026-10-23 21:28:17 +08:00
add and update samples for CUDA 11.5
This commit is contained in:
Rutwik Choughule
18
Samples/FDTD3d/.vscode/c_cpp_properties.json
vendored
Normal file
18
Samples/FDTD3d/.vscode/c_cpp_properties.json
vendored
Normal file
@@ -0,0 +1,18 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "Linux",
|
||||
"includePath": [
|
||||
"${workspaceFolder}/**",
|
||||
"${workspaceFolder}/../../Common"
|
||||
],
|
||||
"defines": [],
|
||||
"compilerPath": "/usr/local/cuda/bin/nvcc",
|
||||
"cStandard": "gnu17",
|
||||
"cppStandard": "gnu++14",
|
||||
"intelliSenseMode": "linux-gcc-x64",
|
||||
"configurationProvider": "ms-vscode.makefile-tools"
|
||||
}
|
||||
],
|
||||
"version": 4
|
||||
}
|
||||
7
Samples/FDTD3d/.vscode/extensions.json
vendored
Normal file
7
Samples/FDTD3d/.vscode/extensions.json
vendored
Normal file
@@ -0,0 +1,7 @@
|
||||
{
|
||||
"recommendations": [
|
||||
"nvidia.nsight-vscode-edition",
|
||||
"ms-vscode.cpptools",
|
||||
"ms-vscode.makefile-tools"
|
||||
]
|
||||
}
|
||||
10
Samples/FDTD3d/.vscode/launch.json
vendored
Normal file
10
Samples/FDTD3d/.vscode/launch.json
vendored
Normal file
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"configurations": [
|
||||
{
|
||||
"name": "CUDA C++: Launch",
|
||||
"type": "cuda-gdb",
|
||||
"request": "launch",
|
||||
"program": "${workspaceFolder}/FDTD3d"
|
||||
}
|
||||
]
|
||||
}
|
||||
15
Samples/FDTD3d/.vscode/tasks.json
vendored
Normal file
15
Samples/FDTD3d/.vscode/tasks.json
vendored
Normal file
@@ -0,0 +1,15 @@
|
||||
{
|
||||
"version": "2.0.0",
|
||||
"tasks": [
|
||||
{
|
||||
"label": "sample",
|
||||
"type": "shell",
|
||||
"command": "make dbg=1",
|
||||
"problemMatcher": ["$nvcc"],
|
||||
"group": {
|
||||
"kind": "build",
|
||||
"isDefault": true
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
20
Samples/FDTD3d/FDTD3d_vs2017.sln
Normal file
20
Samples/FDTD3d/FDTD3d_vs2017.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2017
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "FDTD3d", "FDTD3d_vs2017.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
118
Samples/FDTD3d/FDTD3d_vs2017.vcxproj
Normal file
118
Samples/FDTD3d/FDTD3d_vs2017.vcxproj
Normal file
@@ -0,0 +1,118 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>FDTD3d_vs2017</RootNamespace>
|
||||
<ProjectName>FDTD3d</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(WindowsTargetPlatformVersion)'==''">
|
||||
<LatestTargetPlatformVersion>$([Microsoft.Build.Utilities.ToolLocationHelper]::GetLatestSDKTargetPlatformVersion('Windows', '10.0'))</LatestTargetPlatformVersion>
|
||||
<WindowsTargetPlatformVersion Condition="'$(WindowsTargetPlatformVersion)' == ''">$(LatestTargetPlatformVersion)</WindowsTargetPlatformVersion>
|
||||
<TargetPlatformVersion>$(WindowsTargetPlatformVersion)</TargetPlatformVersion>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup>
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<CharacterSet>MultiByte</CharacterSet>
|
||||
<PlatformToolset>v141</PlatformToolset>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Release'">
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
|
||||
<ImportGroup Label="ExtensionSettings">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.props" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets">
|
||||
<Import Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" />
|
||||
</ImportGroup>
|
||||
<PropertyGroup Label="UserMacros" />
|
||||
<PropertyGroup>
|
||||
<IntDir>$(Platform)/$(Configuration)/</IntDir>
|
||||
<IncludePath>$(IncludePath)</IncludePath>
|
||||
<CodeAnalysisRuleSet>AllRules.ruleset</CodeAnalysisRuleSet>
|
||||
<CodeAnalysisRules />
|
||||
<CodeAnalysisRuleAssemblies />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Platform)'=='x64'">
|
||||
<OutDir>../../bin/win64/$(Configuration)/</OutDir>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup>
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PreprocessorDefinitions>WIN32;_MBCS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>./;$(CudaToolkitDir)/include;../../Common;inc;</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cudart_static.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>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/FDTD3d.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;compute_86,sm_86;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
|
||||
<Include>./;../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
<AdditionalOptions>--threads 0</AdditionalOptions>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<ClCompile>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>Default</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MTd</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="src/FDTD3d.cpp" />
|
||||
<CudaCompile Include="src/FDTD3dGPU.cu" />
|
||||
<ClCompile Include="src/FDTD3dReference.cpp" />
|
||||
<ClInclude Include="inc/FDTD3d.h" />
|
||||
<ClInclude Include="inc/FDTD3dGPU.h" />
|
||||
<ClInclude Include="inc/FDTD3dReference.h" />
|
||||
<None Include="inc/FDTD3dGPUKernel.cuh" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
20
Samples/FDTD3d/FDTD3d_vs2019.sln
Normal file
20
Samples/FDTD3d/FDTD3d_vs2019.sln
Normal file
@@ -0,0 +1,20 @@
|
||||
|
||||
Microsoft Visual Studio Solution File, Format Version 12.00
|
||||
# Visual Studio 2019
|
||||
Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "FDTD3d", "FDTD3d_vs2019.vcxproj", "{997E0757-EA74-4A4E-A0FC-47D8C8831A15}"
|
||||
EndProject
|
||||
Global
|
||||
GlobalSection(SolutionConfigurationPlatforms) = preSolution
|
||||
Debug|x64 = Debug|x64
|
||||
Release|x64 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(ProjectConfigurationPlatforms) = postSolution
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.ActiveCfg = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Debug|x64.Build.0 = Debug|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.ActiveCfg = Release|x64
|
||||
{997E0757-EA74-4A4E-A0FC-47D8C8831A15}.Release|x64.Build.0 = Release|x64
|
||||
EndGlobalSection
|
||||
GlobalSection(SolutionProperties) = preSolution
|
||||
HideSolutionNode = FALSE
|
||||
EndGlobalSection
|
||||
EndGlobal
|
||||
114
Samples/FDTD3d/FDTD3d_vs2019.vcxproj
Normal file
114
Samples/FDTD3d/FDTD3d_vs2019.vcxproj
Normal file
@@ -0,0 +1,114 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<PropertyGroup>
|
||||
<CUDAPropsPath Condition="'$(CUDAPropsPath)'==''">$(VCTargetsPath)\BuildCustomizations</CUDAPropsPath>
|
||||
</PropertyGroup>
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{997E0757-EA74-4A4E-A0FC-47D8C8831A15}</ProjectGuid>
|
||||
<RootNamespace>FDTD3d_vs2019</RootNamespace>
|
||||
<ProjectName>FDTD3d</ProjectName>
|
||||
<CudaToolkitCustomDir />
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup>
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<CharacterSet>MultiByte</CharacterSet>
|
||||
<PlatformToolset>v142</PlatformToolset>
|
||||
<WindowsTargetPlatformVersion>10.0</WindowsTargetPlatformVersion>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)'=='Release'">
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
|
||||
<ImportGroup Label="ExtensionSettings">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.props" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets">
|
||||
<Import Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" />
|
||||
</ImportGroup>
|
||||
<PropertyGroup Label="UserMacros" />
|
||||
<PropertyGroup>
|
||||
<IntDir>$(Platform)/$(Configuration)/</IntDir>
|
||||
<IncludePath>$(IncludePath)</IncludePath>
|
||||
<CodeAnalysisRuleSet>AllRules.ruleset</CodeAnalysisRuleSet>
|
||||
<CodeAnalysisRules />
|
||||
<CodeAnalysisRuleAssemblies />
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Platform)'=='x64'">
|
||||
<OutDir>../../bin/win64/$(Configuration)/</OutDir>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup>
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PreprocessorDefinitions>WIN32;_MBCS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>./;$(CudaToolkitDir)/include;../../Common;inc;</AdditionalIncludeDirectories>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<AdditionalDependencies>cudart_static.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
|
||||
<AdditionalLibraryDirectories>$(CudaToolkitLibDir);</AdditionalLibraryDirectories>
|
||||
<OutputFile>$(OutDir)/FDTD3d.exe</OutputFile>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<CodeGeneration>compute_35,sm_35;compute_37,sm_37;compute_50,sm_50;compute_52,sm_52;compute_60,sm_60;compute_61,sm_61;compute_70,sm_70;compute_75,sm_75;compute_80,sm_80;compute_86,sm_86;</CodeGeneration>
|
||||
<AdditionalOptions>-Xcompiler "/wd 4819" %(AdditionalOptions)</AdditionalOptions>
|
||||
<Include>./;../../Common</Include>
|
||||
<Defines>WIN32</Defines>
|
||||
<AdditionalOptions>--threads 0</AdditionalOptions>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Debug'">
|
||||
<ClCompile>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>Default</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MTd</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)'=='Release'">
|
||||
<ClCompile>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<RuntimeLibrary>MultiThreaded</RuntimeLibrary>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<GenerateDebugInformation>false</GenerateDebugInformation>
|
||||
<LinkTimeCodeGeneration>UseLinkTimeCodeGeneration</LinkTimeCodeGeneration>
|
||||
</Link>
|
||||
<CudaCompile>
|
||||
<Runtime>MT</Runtime>
|
||||
<TargetMachinePlatform>64</TargetMachinePlatform>
|
||||
</CudaCompile>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<ClCompile Include="src/FDTD3d.cpp" />
|
||||
<CudaCompile Include="src/FDTD3dGPU.cu" />
|
||||
<ClCompile Include="src/FDTD3dReference.cpp" />
|
||||
<ClInclude Include="inc/FDTD3d.h" />
|
||||
<ClInclude Include="inc/FDTD3dGPU.h" />
|
||||
<ClInclude Include="inc/FDTD3dReference.h" />
|
||||
<None Include="inc/FDTD3dGPUKernel.cuh" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
<Import Project="$(CUDAPropsPath)\CUDA 11.5.targets" />
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
370
Samples/FDTD3d/Makefile
Normal file
370
Samples/FDTD3d/Makefile
Normal file
@@ -0,0 +1,370 @@
|
||||
################################################################################
|
||||
# Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
|
||||
#
|
||||
# Redistribution and use in source and binary forms, with or without
|
||||
# modification, are permitted provided that the following conditions
|
||||
# are met:
|
||||
# * Redistributions of source code must retain the above copyright
|
||||
# notice, this list of conditions and the following disclaimer.
|
||||
# * Redistributions in binary form must reproduce the above copyright
|
||||
# notice, this list of conditions and the following disclaimer in the
|
||||
# documentation and/or other materials provided with the distribution.
|
||||
# * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
# contributors may be used to endorse or promote products derived
|
||||
# from this software without specific prior written permission.
|
||||
#
|
||||
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
#
|
||||
################################################################################
|
||||
#
|
||||
# Makefile project only supported on Mac OS X and Linux Platforms)
|
||||
#
|
||||
################################################################################
|
||||
|
||||
# Location of the CUDA Toolkit
|
||||
CUDA_PATH ?= /usr/local/cuda
|
||||
|
||||
##############################
|
||||
# start deprecated interface #
|
||||
##############################
|
||||
ifeq ($(x86_64),1)
|
||||
$(info WARNING - x86_64 variable has been deprecated)
|
||||
$(info WARNING - please use TARGET_ARCH=x86_64 instead)
|
||||
TARGET_ARCH ?= x86_64
|
||||
endif
|
||||
ifeq ($(ARMv7),1)
|
||||
$(info WARNING - ARMv7 variable has been deprecated)
|
||||
$(info WARNING - please use TARGET_ARCH=armv7l instead)
|
||||
TARGET_ARCH ?= armv7l
|
||||
endif
|
||||
ifeq ($(aarch64),1)
|
||||
$(info WARNING - aarch64 variable has been deprecated)
|
||||
$(info WARNING - please use TARGET_ARCH=aarch64 instead)
|
||||
TARGET_ARCH ?= aarch64
|
||||
endif
|
||||
ifeq ($(ppc64le),1)
|
||||
$(info WARNING - ppc64le variable has been deprecated)
|
||||
$(info WARNING - please use TARGET_ARCH=ppc64le instead)
|
||||
TARGET_ARCH ?= ppc64le
|
||||
endif
|
||||
ifneq ($(GCC),)
|
||||
$(info WARNING - GCC variable has been deprecated)
|
||||
$(info WARNING - please use HOST_COMPILER=$(GCC) instead)
|
||||
HOST_COMPILER ?= $(GCC)
|
||||
endif
|
||||
ifneq ($(abi),)
|
||||
$(error ERROR - abi variable has been removed)
|
||||
endif
|
||||
############################
|
||||
# end deprecated interface #
|
||||
############################
|
||||
|
||||
# architecture
|
||||
HOST_ARCH := $(shell uname -m)
|
||||
TARGET_ARCH ?= $(HOST_ARCH)
|
||||
ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le armv7l))
|
||||
ifneq ($(TARGET_ARCH),$(HOST_ARCH))
|
||||
ifneq (,$(filter $(TARGET_ARCH),x86_64 aarch64 sbsa ppc64le))
|
||||
TARGET_SIZE := 64
|
||||
else ifneq (,$(filter $(TARGET_ARCH),armv7l))
|
||||
TARGET_SIZE := 32
|
||||
endif
|
||||
else
|
||||
TARGET_SIZE := $(shell getconf LONG_BIT)
|
||||
endif
|
||||
else
|
||||
$(error ERROR - unsupported value $(TARGET_ARCH) for TARGET_ARCH!)
|
||||
endif
|
||||
|
||||
# sbsa and aarch64 systems look similar. Need to differentiate them at host level for now.
|
||||
ifeq ($(HOST_ARCH),aarch64)
|
||||
ifeq ($(CUDA_PATH)/targets/sbsa-linux,$(shell ls -1d $(CUDA_PATH)/targets/sbsa-linux 2>/dev/null))
|
||||
HOST_ARCH := sbsa
|
||||
TARGET_ARCH := sbsa
|
||||
endif
|
||||
endif
|
||||
|
||||
ifneq ($(TARGET_ARCH),$(HOST_ARCH))
|
||||
ifeq (,$(filter $(HOST_ARCH)-$(TARGET_ARCH),aarch64-armv7l x86_64-armv7l x86_64-aarch64 x86_64-sbsa x86_64-ppc64le))
|
||||
$(error ERROR - cross compiling from $(HOST_ARCH) to $(TARGET_ARCH) is not supported!)
|
||||
endif
|
||||
endif
|
||||
|
||||
# When on native aarch64 system with userspace of 32-bit, change TARGET_ARCH to armv7l
|
||||
ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_SIZE),aarch64-aarch64-32)
|
||||
TARGET_ARCH = armv7l
|
||||
endif
|
||||
|
||||
# operating system
|
||||
HOST_OS := $(shell uname -s 2>/dev/null | tr "[:upper:]" "[:lower:]")
|
||||
TARGET_OS ?= $(HOST_OS)
|
||||
ifeq (,$(filter $(TARGET_OS),linux darwin qnx android))
|
||||
$(error ERROR - unsupported value $(TARGET_OS) for TARGET_OS!)
|
||||
endif
|
||||
|
||||
# host compiler
|
||||
ifeq ($(TARGET_OS),darwin)
|
||||
ifeq ($(shell expr `xcodebuild -version | grep -i xcode | awk '{print $$2}' | cut -d'.' -f1` \>= 5),1)
|
||||
HOST_COMPILER ?= clang++
|
||||
endif
|
||||
else ifneq ($(TARGET_ARCH),$(HOST_ARCH))
|
||||
ifeq ($(HOST_ARCH)-$(TARGET_ARCH),x86_64-armv7l)
|
||||
ifeq ($(TARGET_OS),linux)
|
||||
HOST_COMPILER ?= arm-linux-gnueabihf-g++
|
||||
else ifeq ($(TARGET_OS),qnx)
|
||||
ifeq ($(QNX_HOST),)
|
||||
$(error ERROR - QNX_HOST must be passed to the QNX host toolchain)
|
||||
endif
|
||||
ifeq ($(QNX_TARGET),)
|
||||
$(error ERROR - QNX_TARGET must be passed to the QNX target toolchain)
|
||||
endif
|
||||
export QNX_HOST
|
||||
export QNX_TARGET
|
||||
HOST_COMPILER ?= $(QNX_HOST)/usr/bin/arm-unknown-nto-qnx6.6.0eabi-g++
|
||||
else ifeq ($(TARGET_OS),android)
|
||||
HOST_COMPILER ?= arm-linux-androideabi-g++
|
||||
endif
|
||||
else ifeq ($(TARGET_ARCH),aarch64)
|
||||
ifeq ($(TARGET_OS), linux)
|
||||
HOST_COMPILER ?= aarch64-linux-gnu-g++
|
||||
else ifeq ($(TARGET_OS),qnx)
|
||||
ifeq ($(QNX_HOST),)
|
||||
$(error ERROR - QNX_HOST must be passed to the QNX host toolchain)
|
||||
endif
|
||||
ifeq ($(QNX_TARGET),)
|
||||
$(error ERROR - QNX_TARGET must be passed to the QNX target toolchain)
|
||||
endif
|
||||
export QNX_HOST
|
||||
export QNX_TARGET
|
||||
HOST_COMPILER ?= $(QNX_HOST)/usr/bin/q++
|
||||
else ifeq ($(TARGET_OS), android)
|
||||
HOST_COMPILER ?= aarch64-linux-android-clang++
|
||||
endif
|
||||
else ifeq ($(TARGET_ARCH),sbsa)
|
||||
HOST_COMPILER ?= aarch64-linux-gnu-g++
|
||||
else ifeq ($(TARGET_ARCH),ppc64le)
|
||||
HOST_COMPILER ?= powerpc64le-linux-gnu-g++
|
||||
endif
|
||||
endif
|
||||
HOST_COMPILER ?= g++
|
||||
NVCC := $(CUDA_PATH)/bin/nvcc -ccbin $(HOST_COMPILER)
|
||||
|
||||
# internal flags
|
||||
NVCCFLAGS := -m${TARGET_SIZE}
|
||||
CCFLAGS :=
|
||||
LDFLAGS :=
|
||||
|
||||
# build flags
|
||||
ifeq ($(TARGET_OS),darwin)
|
||||
LDFLAGS += -rpath $(CUDA_PATH)/lib
|
||||
CCFLAGS += -arch $(HOST_ARCH)
|
||||
else ifeq ($(HOST_ARCH)-$(TARGET_ARCH)-$(TARGET_OS),x86_64-armv7l-linux)
|
||||
LDFLAGS += --dynamic-linker=/lib/ld-linux-armhf.so.3
|
||||
CCFLAGS += -mfloat-abi=hard
|
||||
else ifeq ($(TARGET_OS),android)
|
||||
LDFLAGS += -pie
|
||||
CCFLAGS += -fpie -fpic -fexceptions
|
||||
endif
|
||||
|
||||
ifneq ($(TARGET_ARCH),$(HOST_ARCH))
|
||||
ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux)
|
||||
ifneq ($(TARGET_FS),)
|
||||
GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6)
|
||||
ifeq ($(GCCVERSIONLTEQ46),1)
|
||||
CCFLAGS += --sysroot=$(TARGET_FS)
|
||||
endif
|
||||
LDFLAGS += --sysroot=$(TARGET_FS)
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/lib
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/arm-linux-gnueabihf
|
||||
endif
|
||||
endif
|
||||
ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux)
|
||||
ifneq ($(TARGET_FS),)
|
||||
GCCVERSIONLTEQ46 := $(shell expr `$(HOST_COMPILER) -dumpversion` \<= 4.6)
|
||||
ifeq ($(GCCVERSIONLTEQ46),1)
|
||||
CCFLAGS += --sysroot=$(TARGET_FS)
|
||||
endif
|
||||
LDFLAGS += --sysroot=$(TARGET_FS)
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/lib -L$(TARGET_FS)/lib
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/lib/aarch64-linux-gnu -L$(TARGET_FS)/lib/aarch64-linux-gnu
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib -L$(TARGET_FS)/usr/lib
|
||||
LDFLAGS += -rpath-link=$(TARGET_FS)/usr/lib/aarch64-linux-gnu -L$(TARGET_FS)/usr/lib/aarch64-linux-gnu
|
||||
LDFLAGS += --unresolved-symbols=ignore-in-shared-libs
|
||||
CCFLAGS += -isystem=$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include -I$(TARGET_FS)/usr/include/libdrm
|
||||
CCFLAGS += -isystem=$(TARGET_FS)/usr/include/aarch64-linux-gnu -I$(TARGET_FS)/usr/include/aarch64-linux-gnu
|
||||
endif
|
||||
endif
|
||||
ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx)
|
||||
NVCCFLAGS += --qpp-config 5.4.0,gcc_ntoaarch64le
|
||||
CCFLAGS += -DWIN_INTERFACE_CUSTOM -I/usr/include/aarch64-qnx-gnu
|
||||
LDFLAGS += -lsocket
|
||||
LDFLAGS += -L/usr/lib/aarch64-qnx-gnu
|
||||
CCFLAGS += "-Wl\,-rpath-link\,/usr/lib/aarch64-qnx-gnu"
|
||||
ifdef TARGET_OVERRIDE
|
||||
LDFLAGS += -lslog2
|
||||
endif
|
||||
|
||||
ifneq ($(TARGET_FS),)
|
||||
LDFLAGS += -L$(TARGET_FS)/usr/lib
|
||||
CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/lib"
|
||||
LDFLAGS += -L$(TARGET_FS)/usr/libnvidia
|
||||
CCFLAGS += "-Wl\,-rpath-link\,$(TARGET_FS)/usr/libnvidia"
|
||||
CCFLAGS += -I$(TARGET_FS)/../include
|
||||
endif
|
||||
endif
|
||||
endif
|
||||
|
||||
ifdef TARGET_OVERRIDE # cuda toolkit targets override
|
||||
NVCCFLAGS += -target-dir $(TARGET_OVERRIDE)
|
||||
endif
|
||||
|
||||
# Install directory of different arch
|
||||
CUDA_INSTALL_TARGET_DIR :=
|
||||
ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-linux)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-gnueabihf/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-linux)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),sbsa-linux)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/sbsa-linux/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-android)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/armv7-linux-androideabi/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-android)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/aarch64-linux-androideabi/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),armv7l-qnx)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/ARMv7-linux-QNX/
|
||||
else ifeq ($(TARGET_ARCH)-$(TARGET_OS),aarch64-qnx)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/aarch64-qnx/
|
||||
else ifeq ($(TARGET_ARCH),ppc64le)
|
||||
CUDA_INSTALL_TARGET_DIR = targets/ppc64le-linux/
|
||||
endif
|
||||
|
||||
# Debug build flags
|
||||
ifeq ($(dbg),1)
|
||||
NVCCFLAGS += -g -G
|
||||
BUILD_TYPE := debug
|
||||
else
|
||||
BUILD_TYPE := release
|
||||
endif
|
||||
|
||||
ALL_CCFLAGS :=
|
||||
ALL_CCFLAGS += $(NVCCFLAGS)
|
||||
ALL_CCFLAGS += $(EXTRA_NVCCFLAGS)
|
||||
ALL_CCFLAGS += $(addprefix -Xcompiler ,$(CCFLAGS))
|
||||
ALL_CCFLAGS += $(addprefix -Xcompiler ,$(EXTRA_CCFLAGS))
|
||||
|
||||
SAMPLE_ENABLED := 1
|
||||
|
||||
ALL_LDFLAGS :=
|
||||
ALL_LDFLAGS += $(ALL_CCFLAGS)
|
||||
ALL_LDFLAGS += $(addprefix -Xlinker ,$(LDFLAGS))
|
||||
ALL_LDFLAGS += $(addprefix -Xlinker ,$(EXTRA_LDFLAGS))
|
||||
|
||||
# Common includes and paths for CUDA
|
||||
INCLUDES := -I../../Common
|
||||
LIBRARIES :=
|
||||
|
||||
################################################################################
|
||||
|
||||
#Detect if installed version of GCC supports required C++11
|
||||
ifeq ($(TARGET_OS),linux)
|
||||
empty :=
|
||||
space := $(empty) $(empty)
|
||||
GCCVERSIONSTRING := $(shell expr `$(HOST_COMPILER) -dumpversion`)
|
||||
#Create version number without "."
|
||||
GCCVERSION := $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f1 -d.)
|
||||
GCCVERSION += $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f2 -d.)
|
||||
GCCVERSION += $(shell expr `echo $(GCCVERSIONSTRING)` | cut -f3 -d.)
|
||||
# Make sure the version number has at least 3 decimals
|
||||
GCCVERSION += 00
|
||||
# Remove spaces from the version number
|
||||
GCCVERSION := $(subst $(space),$(empty),$(GCCVERSION))
|
||||
#$(warning $(GCCVERSION))
|
||||
|
||||
IS_MIN_VERSION := $(shell expr `echo $(GCCVERSION)` \>= 51000)
|
||||
|
||||
ifeq ($(IS_MIN_VERSION), 1)
|
||||
$(info >>> GCC Version is greater or equal to 5.1.0 <<<)
|
||||
else
|
||||
$(info >>> Waiving build. Minimum GCC version required is 5.1.0<<<)
|
||||
SAMPLE_ENABLED := 0
|
||||
endif
|
||||
endif
|
||||
|
||||
# Gencode arguments
|
||||
ifeq ($(TARGET_ARCH),$(filter $(TARGET_ARCH),armv7l aarch64))
|
||||
SMS ?= 35 37 50 52 60 61 70 72 75 80 86
|
||||
else
|
||||
SMS ?= 35 37 50 52 60 61 70 75 80 86
|
||||
endif
|
||||
|
||||
ifeq ($(SMS),)
|
||||
$(info >>> WARNING - no SM architectures have been specified - waiving sample <<<)
|
||||
SAMPLE_ENABLED := 0
|
||||
endif
|
||||
|
||||
ifeq ($(GENCODE_FLAGS),)
|
||||
# Generate SASS code for each SM architecture listed in $(SMS)
|
||||
$(foreach sm,$(SMS),$(eval GENCODE_FLAGS += -gencode arch=compute_$(sm),code=sm_$(sm)))
|
||||
|
||||
# Generate PTX code from the highest SM architecture in $(SMS) to guarantee forward-compatibility
|
||||
HIGHEST_SM := $(lastword $(sort $(SMS)))
|
||||
ifneq ($(HIGHEST_SM),)
|
||||
GENCODE_FLAGS += -gencode arch=compute_$(HIGHEST_SM),code=compute_$(HIGHEST_SM)
|
||||
endif
|
||||
endif
|
||||
|
||||
ALL_CCFLAGS += --threads 0 --std=c++11
|
||||
|
||||
INCLUDES += -Iinc
|
||||
|
||||
ifeq ($(SAMPLE_ENABLED),0)
|
||||
EXEC ?= @echo "[@]"
|
||||
endif
|
||||
|
||||
################################################################################
|
||||
|
||||
# Target rules
|
||||
all: build
|
||||
|
||||
build: FDTD3d
|
||||
|
||||
check.deps:
|
||||
ifeq ($(SAMPLE_ENABLED),0)
|
||||
@echo "Sample will be waived due to the above missing dependencies"
|
||||
else
|
||||
@echo "Sample is ready - all dependencies have been met"
|
||||
endif
|
||||
|
||||
FDTD3d.o:src/FDTD3d.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
FDTD3dGPU.o:src/FDTD3dGPU.cu
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
FDTD3dReference.o:src/FDTD3dReference.cpp
|
||||
$(EXEC) $(NVCC) $(INCLUDES) $(ALL_CCFLAGS) $(GENCODE_FLAGS) -o $@ -c $<
|
||||
|
||||
FDTD3d: FDTD3d.o FDTD3dGPU.o FDTD3dReference.o
|
||||
$(EXEC) $(NVCC) $(ALL_LDFLAGS) $(GENCODE_FLAGS) -o $@ $+ $(LIBRARIES)
|
||||
$(EXEC) mkdir -p ../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)
|
||||
$(EXEC) cp $@ ../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)
|
||||
|
||||
run: build
|
||||
$(EXEC) ./FDTD3d
|
||||
|
||||
clean:
|
||||
rm -f FDTD3d FDTD3d.o FDTD3dGPU.o FDTD3dReference.o
|
||||
rm -rf ../../bin/$(TARGET_ARCH)/$(TARGET_OS)/$(BUILD_TYPE)/FDTD3d
|
||||
|
||||
clobber: clean
|
||||
71
Samples/FDTD3d/NsightEclipse.xml
Normal file
71
Samples/FDTD3d/NsightEclipse.xml
Normal file
@@ -0,0 +1,71 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<!DOCTYPE entry SYSTEM "SamplesInfo.dtd">
|
||||
<entry>
|
||||
<name>FDTD3d</name>
|
||||
<description><![CDATA[This sample applies a finite differences time domain progression stencil on a 3D surface.]]></description>
|
||||
<devicecompilation>whole</devicecompilation>
|
||||
<includepaths>
|
||||
<path>inc</path>
|
||||
<path>./</path>
|
||||
<path>../</path>
|
||||
<path>../../Common</path>
|
||||
</includepaths>
|
||||
<keyconcepts>
|
||||
<concept level="advanced">Performance Strategies</concept>
|
||||
</keyconcepts>
|
||||
<keywords>
|
||||
<keyword>GPGPU</keyword>
|
||||
<keyword>CUDA</keyword>
|
||||
<keyword>finite difference</keyword>
|
||||
<keyword>fdtd</keyword>
|
||||
<keyword>differential equation</keyword>
|
||||
<keyword>pde</keyword>
|
||||
<keyword>ode</keyword>
|
||||
</keywords>
|
||||
<libraries>
|
||||
</libraries>
|
||||
<librarypaths>
|
||||
</librarypaths>
|
||||
<nsight_eclipse>true</nsight_eclipse>
|
||||
<primary_file>FDTD3d.cpp</primary_file>
|
||||
<scopes>
|
||||
<scope>1:CUDA Advanced Topics</scope>
|
||||
<scope>1:Performance Strategies</scope>
|
||||
</scopes>
|
||||
<sm-arch>sm35</sm-arch>
|
||||
<sm-arch>sm37</sm-arch>
|
||||
<sm-arch>sm50</sm-arch>
|
||||
<sm-arch>sm52</sm-arch>
|
||||
<sm-arch>sm60</sm-arch>
|
||||
<sm-arch>sm61</sm-arch>
|
||||
<sm-arch>sm70</sm-arch>
|
||||
<sm-arch>sm72</sm-arch>
|
||||
<sm-arch>sm75</sm-arch>
|
||||
<sm-arch>sm80</sm-arch>
|
||||
<sm-arch>sm86</sm-arch>
|
||||
<supported_envs>
|
||||
<env>
|
||||
<arch>x86_64</arch>
|
||||
<platform>linux</platform>
|
||||
</env>
|
||||
<env>
|
||||
<platform>windows7</platform>
|
||||
</env>
|
||||
<env>
|
||||
<arch>x86_64</arch>
|
||||
<platform>macosx</platform>
|
||||
</env>
|
||||
<env>
|
||||
<arch>arm</arch>
|
||||
</env>
|
||||
<env>
|
||||
<arch>ppc64le</arch>
|
||||
<platform>linux</platform>
|
||||
</env>
|
||||
</supported_envs>
|
||||
<supported_sm_architectures>
|
||||
<include>all</include>
|
||||
</supported_sm_architectures>
|
||||
<title>CUDA C 3D FDTD</title>
|
||||
<type>exe</type>
|
||||
</entry>
|
||||
67
Samples/FDTD3d/README.md
Normal file
67
Samples/FDTD3d/README.md
Normal file
@@ -0,0 +1,67 @@
|
||||
# FDTD3d - CUDA C 3D FDTD
|
||||
|
||||
## Description
|
||||
|
||||
This sample applies a finite differences time domain progression stencil on a 3D surface.
|
||||
|
||||
## Key Concepts
|
||||
|
||||
Performance Strategies
|
||||
|
||||
## Supported SM Architectures
|
||||
|
||||
[SM 3.5 ](https://developer.nvidia.com/cuda-gpus) [SM 3.7 ](https://developer.nvidia.com/cuda-gpus) [SM 5.0 ](https://developer.nvidia.com/cuda-gpus) [SM 5.2 ](https://developer.nvidia.com/cuda-gpus) [SM 6.0 ](https://developer.nvidia.com/cuda-gpus) [SM 6.1 ](https://developer.nvidia.com/cuda-gpus) [SM 7.0 ](https://developer.nvidia.com/cuda-gpus) [SM 7.2 ](https://developer.nvidia.com/cuda-gpus) [SM 7.5 ](https://developer.nvidia.com/cuda-gpus) [SM 8.0 ](https://developer.nvidia.com/cuda-gpus) [SM 8.6 ](https://developer.nvidia.com/cuda-gpus)
|
||||
|
||||
## Supported OSes
|
||||
|
||||
Linux, Windows
|
||||
|
||||
## Supported CPU Architecture
|
||||
|
||||
x86_64, ppc64le, armv7l
|
||||
|
||||
## CUDA APIs involved
|
||||
|
||||
## Prerequisites
|
||||
|
||||
Download and install the [CUDA Toolkit 11.5](https://developer.nvidia.com/cuda-downloads) for your corresponding platform.
|
||||
|
||||
## Build and Run
|
||||
|
||||
### Windows
|
||||
The Windows samples are built using the Visual Studio IDE. Solution files (.sln) are provided for each supported version of Visual Studio, using the format:
|
||||
```
|
||||
*_vs<version>.sln - for Visual Studio <version>
|
||||
```
|
||||
Each individual sample has its own set of solution files in its directory:
|
||||
|
||||
To build/examine all the samples at once, the complete solution files should be used. To build/examine a single sample, the individual sample solution files should be used.
|
||||
> **Note:** Some samples require that the Microsoft DirectX SDK (June 2010 or newer) be installed and that the VC++ directory paths are properly set up (**Tools > Options...**). Check DirectX Dependencies section for details."
|
||||
|
||||
### Linux
|
||||
The Linux samples are built using makefiles. To use the makefiles, change the current directory to the sample directory you wish to build, and run make:
|
||||
```
|
||||
$ cd <sample_dir>
|
||||
$ make
|
||||
```
|
||||
The samples makefiles can take advantage of certain options:
|
||||
* **TARGET_ARCH=<arch>** - cross-compile targeting a specific architecture. Allowed architectures are x86_64, ppc64le, armv7l.
|
||||
By default, TARGET_ARCH is set to HOST_ARCH. On a x86_64 machine, not setting TARGET_ARCH is the equivalent of setting TARGET_ARCH=x86_64.<br/>
|
||||
`$ make TARGET_ARCH=x86_64` <br/> `$ make TARGET_ARCH=ppc64le` <br/> `$ make TARGET_ARCH=armv7l` <br/>
|
||||
See [here](http://docs.nvidia.com/cuda/cuda-samples/index.html#cross-samples) for more details.
|
||||
* **dbg=1** - build with debug symbols
|
||||
```
|
||||
$ make dbg=1
|
||||
```
|
||||
* **SMS="A B ..."** - override the SM architectures for which the sample will be built, where `"A B ..."` is a space-delimited list of SM architectures. For example, to generate SASS for SM 50 and SM 60, use `SMS="50 60"`.
|
||||
```
|
||||
$ make SMS="50 60"
|
||||
```
|
||||
|
||||
* **HOST_COMPILER=<host_compiler>** - override the default g++ host compiler. See the [Linux Installation Guide](http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements) for a list of supported host compilers.
|
||||
```
|
||||
$ make HOST_COMPILER=g++
|
||||
```
|
||||
|
||||
## References (for more details)
|
||||
|
||||
54
Samples/FDTD3d/inc/FDTD3d.h
Normal file
54
Samples/FDTD3d/inc/FDTD3d.h
Normal file
@@ -0,0 +1,54 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#ifndef _FDTD3D_H_
|
||||
#define _FDTD3D_H_
|
||||
|
||||
// The values are set to give reasonable runtimes, they can
|
||||
// be changed but note that running very large dimensions can
|
||||
// take a very long time and you should avoid running on your
|
||||
// primary display in this case.
|
||||
#define k_dim_min 96
|
||||
#define k_dim_max 376
|
||||
#define k_dim_qa 248
|
||||
|
||||
// Note that the radius is defined here as exactly 4 since the
|
||||
// kernel code uses a constant. If you want a different radius
|
||||
// you must change the kernel accordingly.
|
||||
#define k_radius_min 4
|
||||
#define k_radius_max 4
|
||||
#define k_radius_default 4
|
||||
|
||||
// The values are set to give reasonable runtimes, they can
|
||||
// be changed but note that running a very large number of
|
||||
// timesteps can take a very long time and you should avoid
|
||||
// running on your primary display in this case.
|
||||
#define k_timesteps_min 1
|
||||
#define k_timesteps_max 10
|
||||
#define k_timesteps_default 5
|
||||
|
||||
#endif
|
||||
51
Samples/FDTD3d/inc/FDTD3dGPU.h
Normal file
51
Samples/FDTD3d/inc/FDTD3dGPU.h
Normal file
@@ -0,0 +1,51 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#ifndef _FDTD3DGPU_H_
|
||||
#define _FDTD3DGPU_H_
|
||||
|
||||
#include <cstddef>
|
||||
#if defined(WIN32) || defined(_WIN32) || defined(WIN64) || \
|
||||
defined(_WIN64) && defined(_MSC_VER)
|
||||
typedef unsigned __int64 memsize_t;
|
||||
#else
|
||||
#include <stdint.h>
|
||||
typedef uint64_t memsize_t;
|
||||
#endif
|
||||
|
||||
#define k_blockDimX 32
|
||||
#define k_blockDimMaxY 16
|
||||
#define k_blockSizeMin 128
|
||||
#define k_blockSizeMax (k_blockDimX * k_blockDimMaxY)
|
||||
|
||||
bool getTargetDeviceGlobalMemSize(memsize_t *result, const int argc,
|
||||
const char **argv);
|
||||
bool fdtdGPU(float *output, const float *input, const float *coeff,
|
||||
const int dimx, const int dimy, const int dimz, const int radius,
|
||||
const int timesteps, const int argc, const char **argv);
|
||||
|
||||
#endif
|
||||
150
Samples/FDTD3d/inc/FDTD3dGPUKernel.cuh
Normal file
150
Samples/FDTD3d/inc/FDTD3dGPUKernel.cuh
Normal file
@@ -0,0 +1,150 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#include "FDTD3dGPU.h"
|
||||
#include <cooperative_groups.h>
|
||||
|
||||
namespace cg = cooperative_groups;
|
||||
|
||||
// Note: If you change the RADIUS, you should also change the unrolling below
|
||||
#define RADIUS 4
|
||||
|
||||
__constant__ float stencil[RADIUS + 1];
|
||||
|
||||
__global__ void FiniteDifferencesKernel(float *output, const float *input,
|
||||
const int dimx, const int dimy,
|
||||
const int dimz) {
|
||||
bool validr = true;
|
||||
bool validw = true;
|
||||
const int gtidx = blockIdx.x * blockDim.x + threadIdx.x;
|
||||
const int gtidy = blockIdx.y * blockDim.y + threadIdx.y;
|
||||
const int ltidx = threadIdx.x;
|
||||
const int ltidy = threadIdx.y;
|
||||
const int workx = blockDim.x;
|
||||
const int worky = blockDim.y;
|
||||
// Handle to thread block group
|
||||
cg::thread_block cta = cg::this_thread_block();
|
||||
__shared__ float tile[k_blockDimMaxY + 2 * RADIUS][k_blockDimX + 2 * RADIUS];
|
||||
|
||||
const int stride_y = dimx + 2 * RADIUS;
|
||||
const int stride_z = stride_y * (dimy + 2 * RADIUS);
|
||||
|
||||
int inputIndex = 0;
|
||||
int outputIndex = 0;
|
||||
|
||||
// Advance inputIndex to start of inner volume
|
||||
inputIndex += RADIUS * stride_y + RADIUS;
|
||||
|
||||
// Advance inputIndex to target element
|
||||
inputIndex += gtidy * stride_y + gtidx;
|
||||
|
||||
float infront[RADIUS];
|
||||
float behind[RADIUS];
|
||||
float current;
|
||||
|
||||
const int tx = ltidx + RADIUS;
|
||||
const int ty = ltidy + RADIUS;
|
||||
|
||||
// Check in bounds
|
||||
if ((gtidx >= dimx + RADIUS) || (gtidy >= dimy + RADIUS)) validr = false;
|
||||
|
||||
if ((gtidx >= dimx) || (gtidy >= dimy)) validw = false;
|
||||
|
||||
// Preload the "infront" and "behind" data
|
||||
for (int i = RADIUS - 2; i >= 0; i--) {
|
||||
if (validr) behind[i] = input[inputIndex];
|
||||
|
||||
inputIndex += stride_z;
|
||||
}
|
||||
|
||||
if (validr) current = input[inputIndex];
|
||||
|
||||
outputIndex = inputIndex;
|
||||
inputIndex += stride_z;
|
||||
|
||||
for (int i = 0; i < RADIUS; i++) {
|
||||
if (validr) infront[i] = input[inputIndex];
|
||||
|
||||
inputIndex += stride_z;
|
||||
}
|
||||
|
||||
// Step through the xy-planes
|
||||
#pragma unroll 9
|
||||
|
||||
for (int iz = 0; iz < dimz; iz++) {
|
||||
// Advance the slice (move the thread-front)
|
||||
for (int i = RADIUS - 1; i > 0; i--) behind[i] = behind[i - 1];
|
||||
|
||||
behind[0] = current;
|
||||
current = infront[0];
|
||||
#pragma unroll 4
|
||||
|
||||
for (int i = 0; i < RADIUS - 1; i++) infront[i] = infront[i + 1];
|
||||
|
||||
if (validr) infront[RADIUS - 1] = input[inputIndex];
|
||||
|
||||
inputIndex += stride_z;
|
||||
outputIndex += stride_z;
|
||||
cg::sync(cta);
|
||||
|
||||
// Note that for the work items on the boundary of the problem, the
|
||||
// supplied index when reading the halo (below) may wrap to the
|
||||
// previous/next row or even the previous/next xy-plane. This is
|
||||
// acceptable since a) we disable the output write for these work
|
||||
// items and b) there is at least one xy-plane before/after the
|
||||
// current plane, so the access will be within bounds.
|
||||
|
||||
// Update the data slice in the local tile
|
||||
// Halo above & below
|
||||
if (ltidy < RADIUS) {
|
||||
tile[ltidy][tx] = input[outputIndex - RADIUS * stride_y];
|
||||
tile[ltidy + worky + RADIUS][tx] = input[outputIndex + worky * stride_y];
|
||||
}
|
||||
|
||||
// Halo left & right
|
||||
if (ltidx < RADIUS) {
|
||||
tile[ty][ltidx] = input[outputIndex - RADIUS];
|
||||
tile[ty][ltidx + workx + RADIUS] = input[outputIndex + workx];
|
||||
}
|
||||
|
||||
tile[ty][tx] = current;
|
||||
cg::sync(cta);
|
||||
|
||||
// Compute the output value
|
||||
float value = stencil[0] * current;
|
||||
#pragma unroll 4
|
||||
|
||||
for (int i = 1; i <= RADIUS; i++) {
|
||||
value +=
|
||||
stencil[i] * (infront[i - 1] + behind[i - 1] + tile[ty - i][tx] +
|
||||
tile[ty + i][tx] + tile[ty][tx - i] + tile[ty][tx + i]);
|
||||
}
|
||||
|
||||
// Store the output value
|
||||
if (validw) output[outputIndex] = value;
|
||||
}
|
||||
}
|
||||
44
Samples/FDTD3d/inc/FDTD3dReference.h
Normal file
44
Samples/FDTD3d/inc/FDTD3dReference.h
Normal file
@@ -0,0 +1,44 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#ifndef _FDTD3DREFERENCE_H_
|
||||
#define _FDTD3DREFERENCE_H_
|
||||
|
||||
void generateRandomData(float *data, const int dimx, const int dimy,
|
||||
const int dimz, const float lowerBound,
|
||||
const float upperBound);
|
||||
void generatePatternData(float *data, const int dimx, const int dimy,
|
||||
const int dimz, const float lowerBound,
|
||||
const float upperBound);
|
||||
bool fdtdReference(float *output, const float *input, const float *coeff,
|
||||
const int dimx, const int dimy, const int dimz,
|
||||
const int radius, const int timesteps);
|
||||
bool compareData(const float *output, const float *reference, const int dimx,
|
||||
const int dimy, const int dimz, const int radius,
|
||||
const float tolerance = 0.0001f);
|
||||
|
||||
#endif
|
||||
235
Samples/FDTD3d/src/FDTD3d.cpp
Normal file
235
Samples/FDTD3d/src/FDTD3d.cpp
Normal file
@@ -0,0 +1,235 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#include "FDTD3d.h"
|
||||
|
||||
#include <iostream>
|
||||
#include <iomanip>
|
||||
|
||||
#include "FDTD3dReference.h"
|
||||
#include "FDTD3dGPU.h"
|
||||
|
||||
#include <helper_functions.h>
|
||||
|
||||
#include <math.h>
|
||||
#include <assert.h>
|
||||
|
||||
#ifndef CLAMP
|
||||
#define CLAMP(a, min, max) (MIN(max, MAX(a, min)))
|
||||
#endif
|
||||
|
||||
//// Name of the log file
|
||||
// const char *printfFile = "FDTD3d.txt";
|
||||
|
||||
// Forward declarations
|
||||
bool runTest(int argc, const char **argv);
|
||||
void showHelp(const int argc, const char **argv);
|
||||
|
||||
int main(int argc, char **argv) {
|
||||
bool bTestResult = false;
|
||||
// Start the log
|
||||
printf("%s Starting...\n\n", argv[0]);
|
||||
|
||||
// Check help flag
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "help")) {
|
||||
printf("Displaying help on console\n");
|
||||
showHelp(argc, (const char **)argv);
|
||||
bTestResult = true;
|
||||
} else {
|
||||
// Execute
|
||||
bTestResult = runTest(argc, (const char **)argv);
|
||||
}
|
||||
|
||||
// Finish
|
||||
exit(bTestResult ? EXIT_SUCCESS : EXIT_FAILURE);
|
||||
}
|
||||
|
||||
void showHelp(const int argc, const char **argv) {
|
||||
if (argc > 0) std::cout << std::endl << argv[0] << std::endl;
|
||||
|
||||
std::cout << std::endl << "Syntax:" << std::endl;
|
||||
std::cout << std::left;
|
||||
std::cout << " " << std::setw(20) << "--device=<device>"
|
||||
<< "Specify device to use for execution" << std::endl;
|
||||
std::cout << " " << std::setw(20) << "--dimx=<N>"
|
||||
<< "Specify number of elements in x direction (excluding halo)"
|
||||
<< std::endl;
|
||||
std::cout << " " << std::setw(20) << "--dimy=<N>"
|
||||
<< "Specify number of elements in y direction (excluding halo)"
|
||||
<< std::endl;
|
||||
std::cout << " " << std::setw(20) << "--dimz=<N>"
|
||||
<< "Specify number of elements in z direction (excluding halo)"
|
||||
<< std::endl;
|
||||
std::cout << " " << std::setw(20) << "--radius=<N>"
|
||||
<< "Specify radius of stencil" << std::endl;
|
||||
std::cout << " " << std::setw(20) << "--timesteps=<N>"
|
||||
<< "Specify number of timesteps" << std::endl;
|
||||
std::cout << " " << std::setw(20) << "--block-size=<N>"
|
||||
<< "Specify number of threads per block" << std::endl;
|
||||
std::cout << std::endl;
|
||||
std::cout << " " << std::setw(20) << "--noprompt"
|
||||
<< "Skip prompt before exit" << std::endl;
|
||||
std::cout << std::endl;
|
||||
}
|
||||
|
||||
bool runTest(int argc, const char **argv) {
|
||||
float *host_output;
|
||||
float *device_output;
|
||||
float *input;
|
||||
float *coeff;
|
||||
|
||||
int defaultDim;
|
||||
int dimx;
|
||||
int dimy;
|
||||
int dimz;
|
||||
int outerDimx;
|
||||
int outerDimy;
|
||||
int outerDimz;
|
||||
int radius;
|
||||
int timesteps;
|
||||
size_t volumeSize;
|
||||
memsize_t memsize;
|
||||
|
||||
const float lowerBound = 0.0f;
|
||||
const float upperBound = 1.0f;
|
||||
|
||||
// Determine default dimensions
|
||||
printf("Set-up, based upon target device GMEM size...\n");
|
||||
// Get the memory size of the target device
|
||||
printf(" getTargetDeviceGlobalMemSize\n");
|
||||
getTargetDeviceGlobalMemSize(&memsize, argc, argv);
|
||||
|
||||
// We can never use all the memory so to keep things simple we aim to
|
||||
// use around half the total memory
|
||||
memsize /= 2;
|
||||
|
||||
// Most of our memory use is taken up by the input and output buffers -
|
||||
// two buffers of equal size - and for simplicity the volume is a cube:
|
||||
// dim = floor( (N/2)^(1/3) )
|
||||
defaultDim = (int)floor(pow((memsize / (2.0 * sizeof(float))), 1.0 / 3.0));
|
||||
|
||||
// By default, make the volume edge size an integer multiple of 128B to
|
||||
// improve performance by coalescing memory accesses, in a real
|
||||
// application it would make sense to pad the lines accordingly
|
||||
int roundTarget = 128 / sizeof(float);
|
||||
defaultDim = defaultDim / roundTarget * roundTarget;
|
||||
defaultDim -= k_radius_default * 2;
|
||||
|
||||
// Check dimension is valid
|
||||
if (defaultDim < k_dim_min) {
|
||||
printf(
|
||||
"insufficient device memory (maximum volume on device is %d, must be "
|
||||
"between %d and %d).\n",
|
||||
defaultDim, k_dim_min, k_dim_max);
|
||||
exit(EXIT_FAILURE);
|
||||
} else if (defaultDim > k_dim_max) {
|
||||
defaultDim = k_dim_max;
|
||||
}
|
||||
|
||||
// For QA testing, override default volume size
|
||||
if (checkCmdLineFlag(argc, argv, "qatest")) {
|
||||
defaultDim = MIN(defaultDim, k_dim_qa);
|
||||
}
|
||||
|
||||
// set default dim
|
||||
dimx = defaultDim;
|
||||
dimy = defaultDim;
|
||||
dimz = defaultDim;
|
||||
radius = k_radius_default;
|
||||
timesteps = k_timesteps_default;
|
||||
|
||||
// Parse command line arguments
|
||||
if (checkCmdLineFlag(argc, argv, "dimx")) {
|
||||
dimx =
|
||||
CLAMP(getCmdLineArgumentInt(argc, argv, "dimx"), k_dim_min, k_dim_max);
|
||||
}
|
||||
|
||||
if (checkCmdLineFlag(argc, argv, "dimy")) {
|
||||
dimy =
|
||||
CLAMP(getCmdLineArgumentInt(argc, argv, "dimy"), k_dim_min, k_dim_max);
|
||||
}
|
||||
|
||||
if (checkCmdLineFlag(argc, argv, "dimz")) {
|
||||
dimz =
|
||||
CLAMP(getCmdLineArgumentInt(argc, argv, "dimz"), k_dim_min, k_dim_max);
|
||||
}
|
||||
|
||||
if (checkCmdLineFlag(argc, argv, "radius")) {
|
||||
radius = CLAMP(getCmdLineArgumentInt(argc, argv, "radius"), k_radius_min,
|
||||
k_radius_max);
|
||||
}
|
||||
|
||||
if (checkCmdLineFlag(argc, argv, "timesteps")) {
|
||||
timesteps = CLAMP(getCmdLineArgumentInt(argc, argv, "timesteps"),
|
||||
k_timesteps_min, k_timesteps_max);
|
||||
}
|
||||
|
||||
// Determine volume size
|
||||
outerDimx = dimx + 2 * radius;
|
||||
outerDimy = dimy + 2 * radius;
|
||||
outerDimz = dimz + 2 * radius;
|
||||
volumeSize = outerDimx * outerDimy * outerDimz;
|
||||
|
||||
// Allocate memory
|
||||
host_output = (float *)calloc(volumeSize, sizeof(float));
|
||||
input = (float *)malloc(volumeSize * sizeof(float));
|
||||
coeff = (float *)malloc((radius + 1) * sizeof(float));
|
||||
|
||||
// Create coefficients
|
||||
for (int i = 0; i <= radius; i++) {
|
||||
coeff[i] = 0.1f;
|
||||
}
|
||||
|
||||
// Generate data
|
||||
printf(" generateRandomData\n\n");
|
||||
generateRandomData(input, outerDimx, outerDimy, outerDimz, lowerBound,
|
||||
upperBound);
|
||||
printf(
|
||||
"FDTD on %d x %d x %d volume with symmetric filter radius %d for %d "
|
||||
"timesteps...\n\n",
|
||||
dimx, dimy, dimz, radius, timesteps);
|
||||
|
||||
// Execute on the host
|
||||
printf("fdtdReference...\n");
|
||||
fdtdReference(host_output, input, coeff, dimx, dimy, dimz, radius, timesteps);
|
||||
printf("fdtdReference complete\n");
|
||||
|
||||
// Allocate memory
|
||||
device_output = (float *)calloc(volumeSize, sizeof(float));
|
||||
|
||||
// Execute on the device
|
||||
printf("fdtdGPU...\n");
|
||||
fdtdGPU(device_output, input, coeff, dimx, dimy, dimz, radius, timesteps,
|
||||
argc, argv);
|
||||
printf("fdtdGPU complete\n");
|
||||
|
||||
// Compare the results
|
||||
float tolerance = 0.0001f;
|
||||
printf("\nCompareData (tolerance %f)...\n", tolerance);
|
||||
return compareData(device_output, host_output, dimx, dimy, dimz, radius,
|
||||
tolerance);
|
||||
}
|
||||
263
Samples/FDTD3d/src/FDTD3dGPU.cu
Normal file
263
Samples/FDTD3d/src/FDTD3dGPU.cu
Normal file
@@ -0,0 +1,263 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#include "FDTD3dGPU.h"
|
||||
|
||||
#include <iostream>
|
||||
#include <algorithm>
|
||||
#include <helper_functions.h>
|
||||
#include <helper_cuda.h>
|
||||
|
||||
#include "FDTD3dGPUKernel.cuh"
|
||||
|
||||
bool getTargetDeviceGlobalMemSize(memsize_t *result, const int argc,
|
||||
const char **argv) {
|
||||
int deviceCount = 0;
|
||||
int targetDevice = 0;
|
||||
size_t memsize = 0;
|
||||
|
||||
// Get the number of CUDA enabled GPU devices
|
||||
printf(" cudaGetDeviceCount\n");
|
||||
checkCudaErrors(cudaGetDeviceCount(&deviceCount));
|
||||
|
||||
// Select target device (device 0 by default)
|
||||
targetDevice = findCudaDevice(argc, (const char **)argv);
|
||||
|
||||
// Query target device for maximum memory allocation
|
||||
printf(" cudaGetDeviceProperties\n");
|
||||
struct cudaDeviceProp deviceProp;
|
||||
checkCudaErrors(cudaGetDeviceProperties(&deviceProp, targetDevice));
|
||||
|
||||
memsize = deviceProp.totalGlobalMem;
|
||||
|
||||
// Save the result
|
||||
*result = (memsize_t)memsize;
|
||||
return true;
|
||||
}
|
||||
|
||||
bool fdtdGPU(float *output, const float *input, const float *coeff,
|
||||
const int dimx, const int dimy, const int dimz, const int radius,
|
||||
const int timesteps, const int argc, const char **argv) {
|
||||
const int outerDimx = dimx + 2 * radius;
|
||||
const int outerDimy = dimy + 2 * radius;
|
||||
const int outerDimz = dimz + 2 * radius;
|
||||
const size_t volumeSize = outerDimx * outerDimy * outerDimz;
|
||||
int deviceCount = 0;
|
||||
int targetDevice = 0;
|
||||
float *bufferOut = 0;
|
||||
float *bufferIn = 0;
|
||||
dim3 dimBlock;
|
||||
dim3 dimGrid;
|
||||
|
||||
// Ensure that the inner data starts on a 128B boundary
|
||||
const int padding = (128 / sizeof(float)) - radius;
|
||||
const size_t paddedVolumeSize = volumeSize + padding;
|
||||
|
||||
#ifdef GPU_PROFILING
|
||||
cudaEvent_t profileStart = 0;
|
||||
cudaEvent_t profileEnd = 0;
|
||||
const int profileTimesteps = timesteps - 1;
|
||||
|
||||
if (profileTimesteps < 1) {
|
||||
printf(
|
||||
" cannot profile with fewer than two timesteps (timesteps=%d), "
|
||||
"profiling is disabled.\n",
|
||||
timesteps);
|
||||
}
|
||||
|
||||
#endif
|
||||
|
||||
// Check the radius is valid
|
||||
if (radius != RADIUS) {
|
||||
printf("radius is invalid, must be %d - see kernel for details.\n", RADIUS);
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
// Get the number of CUDA enabled GPU devices
|
||||
checkCudaErrors(cudaGetDeviceCount(&deviceCount));
|
||||
|
||||
// Select target device (device 0 by default)
|
||||
targetDevice = findCudaDevice(argc, (const char **)argv);
|
||||
|
||||
checkCudaErrors(cudaSetDevice(targetDevice));
|
||||
|
||||
// Allocate memory buffers
|
||||
checkCudaErrors(
|
||||
cudaMalloc((void **)&bufferOut, paddedVolumeSize * sizeof(float)));
|
||||
checkCudaErrors(
|
||||
cudaMalloc((void **)&bufferIn, paddedVolumeSize * sizeof(float)));
|
||||
|
||||
// Check for a command-line specified block size
|
||||
int userBlockSize;
|
||||
|
||||
if (checkCmdLineFlag(argc, (const char **)argv, "block-size")) {
|
||||
userBlockSize = getCmdLineArgumentInt(argc, argv, "block-size");
|
||||
// Constrain to a multiple of k_blockDimX
|
||||
userBlockSize = (userBlockSize / k_blockDimX * k_blockDimX);
|
||||
|
||||
// Constrain within allowed bounds
|
||||
userBlockSize = MIN(MAX(userBlockSize, k_blockSizeMin), k_blockSizeMax);
|
||||
} else {
|
||||
userBlockSize = k_blockSizeMax;
|
||||
}
|
||||
|
||||
// Check the device limit on the number of threads
|
||||
struct cudaFuncAttributes funcAttrib;
|
||||
checkCudaErrors(cudaFuncGetAttributes(&funcAttrib, FiniteDifferencesKernel));
|
||||
|
||||
userBlockSize = MIN(userBlockSize, funcAttrib.maxThreadsPerBlock);
|
||||
|
||||
// Set the block size
|
||||
dimBlock.x = k_blockDimX;
|
||||
// Visual Studio 2005 does not like std::min
|
||||
// dimBlock.y = std::min<size_t>(userBlockSize / k_blockDimX,
|
||||
// (size_t)k_blockDimMaxY);
|
||||
dimBlock.y = ((userBlockSize / k_blockDimX) < (size_t)k_blockDimMaxY)
|
||||
? (userBlockSize / k_blockDimX)
|
||||
: (size_t)k_blockDimMaxY;
|
||||
dimGrid.x = (unsigned int)ceil((float)dimx / dimBlock.x);
|
||||
dimGrid.y = (unsigned int)ceil((float)dimy / dimBlock.y);
|
||||
printf(" set block size to %dx%d\n", dimBlock.x, dimBlock.y);
|
||||
printf(" set grid size to %dx%d\n", dimGrid.x, dimGrid.y);
|
||||
|
||||
// Check the block size is valid
|
||||
if (dimBlock.x < RADIUS || dimBlock.y < RADIUS) {
|
||||
printf("invalid block size, x (%d) and y (%d) must be >= radius (%d).\n",
|
||||
dimBlock.x, dimBlock.y, RADIUS);
|
||||
exit(EXIT_FAILURE);
|
||||
}
|
||||
|
||||
// Copy the input to the device input buffer
|
||||
checkCudaErrors(cudaMemcpy(bufferIn + padding, input,
|
||||
volumeSize * sizeof(float),
|
||||
cudaMemcpyHostToDevice));
|
||||
|
||||
// Copy the input to the device output buffer (actually only need the halo)
|
||||
checkCudaErrors(cudaMemcpy(bufferOut + padding, input,
|
||||
volumeSize * sizeof(float),
|
||||
cudaMemcpyHostToDevice));
|
||||
|
||||
// Copy the coefficients to the device coefficient buffer
|
||||
checkCudaErrors(
|
||||
cudaMemcpyToSymbol(stencil, (void *)coeff, (radius + 1) * sizeof(float)));
|
||||
|
||||
#ifdef GPU_PROFILING
|
||||
|
||||
// Create the events
|
||||
checkCudaErrors(cudaEventCreate(&profileStart));
|
||||
checkCudaErrors(cudaEventCreate(&profileEnd));
|
||||
|
||||
#endif
|
||||
|
||||
// Execute the FDTD
|
||||
float *bufferSrc = bufferIn + padding;
|
||||
float *bufferDst = bufferOut + padding;
|
||||
printf(" GPU FDTD loop\n");
|
||||
|
||||
#ifdef GPU_PROFILING
|
||||
// Enqueue start event
|
||||
checkCudaErrors(cudaEventRecord(profileStart, 0));
|
||||
#endif
|
||||
|
||||
for (int it = 0; it < timesteps; it++) {
|
||||
printf("\tt = %d ", it);
|
||||
|
||||
// Launch the kernel
|
||||
printf("launch kernel\n");
|
||||
FiniteDifferencesKernel<<<dimGrid, dimBlock>>>(bufferDst, bufferSrc, dimx,
|
||||
dimy, dimz);
|
||||
|
||||
// Toggle the buffers
|
||||
// Visual Studio 2005 does not like std::swap
|
||||
// std::swap<float *>(bufferSrc, bufferDst);
|
||||
float *tmp = bufferDst;
|
||||
bufferDst = bufferSrc;
|
||||
bufferSrc = tmp;
|
||||
}
|
||||
|
||||
printf("\n");
|
||||
|
||||
#ifdef GPU_PROFILING
|
||||
// Enqueue end event
|
||||
checkCudaErrors(cudaEventRecord(profileEnd, 0));
|
||||
#endif
|
||||
|
||||
// Wait for the kernel to complete
|
||||
checkCudaErrors(cudaDeviceSynchronize());
|
||||
|
||||
// Read the result back, result is in bufferSrc (after final toggle)
|
||||
checkCudaErrors(cudaMemcpy(output, bufferSrc, volumeSize * sizeof(float),
|
||||
cudaMemcpyDeviceToHost));
|
||||
|
||||
// Report time
|
||||
#ifdef GPU_PROFILING
|
||||
float elapsedTimeMS = 0;
|
||||
|
||||
if (profileTimesteps > 0) {
|
||||
checkCudaErrors(
|
||||
cudaEventElapsedTime(&elapsedTimeMS, profileStart, profileEnd));
|
||||
}
|
||||
|
||||
if (profileTimesteps > 0) {
|
||||
// Convert milliseconds to seconds
|
||||
double elapsedTime = elapsedTimeMS * 1.0e-3;
|
||||
double avgElapsedTime = elapsedTime / (double)profileTimesteps;
|
||||
// Determine number of computations per timestep
|
||||
size_t pointsComputed = dimx * dimy * dimz;
|
||||
// Determine throughput
|
||||
double throughputM = 1.0e-6 * (double)pointsComputed / avgElapsedTime;
|
||||
printf(
|
||||
"FDTD3d, Throughput = %.4f MPoints/s, Time = %.5f s, Size = %u Points, "
|
||||
"NumDevsUsed = %u, Blocksize = %u\n",
|
||||
throughputM, avgElapsedTime, pointsComputed, 1,
|
||||
dimBlock.x * dimBlock.y);
|
||||
}
|
||||
|
||||
#endif
|
||||
|
||||
// Cleanup
|
||||
if (bufferIn) {
|
||||
checkCudaErrors(cudaFree(bufferIn));
|
||||
}
|
||||
|
||||
if (bufferOut) {
|
||||
checkCudaErrors(cudaFree(bufferOut));
|
||||
}
|
||||
|
||||
#ifdef GPU_PROFILING
|
||||
|
||||
if (profileStart) {
|
||||
checkCudaErrors(cudaEventDestroy(profileStart));
|
||||
}
|
||||
|
||||
if (profileEnd) {
|
||||
checkCudaErrors(cudaEventDestroy(profileEnd));
|
||||
}
|
||||
|
||||
#endif
|
||||
return true;
|
||||
}
|
||||
178
Samples/FDTD3d/src/FDTD3dReference.cpp
Normal file
178
Samples/FDTD3d/src/FDTD3dReference.cpp
Normal file
@@ -0,0 +1,178 @@
|
||||
/* Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions
|
||||
* are met:
|
||||
* * Redistributions of source code must retain the above copyright
|
||||
* notice, this list of conditions and the following disclaimer.
|
||||
* * Redistributions in binary form must reproduce the above copyright
|
||||
* notice, this list of conditions and the following disclaimer in the
|
||||
* documentation and/or other materials provided with the distribution.
|
||||
* * Neither the name of NVIDIA CORPORATION nor the names of its
|
||||
* contributors may be used to endorse or promote products derived
|
||||
* from this software without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
|
||||
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
||||
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
|
||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
||||
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#include "FDTD3dReference.h"
|
||||
|
||||
#include <cstdlib>
|
||||
#include <cmath>
|
||||
#include <iostream>
|
||||
#include <iomanip>
|
||||
#include <stdio.h>
|
||||
|
||||
void generateRandomData(float *data, const int dimx, const int dimy,
|
||||
const int dimz, const float lowerBound,
|
||||
const float upperBound) {
|
||||
srand(0);
|
||||
|
||||
for (int iz = 0; iz < dimz; iz++) {
|
||||
for (int iy = 0; iy < dimy; iy++) {
|
||||
for (int ix = 0; ix < dimx; ix++) {
|
||||
*data = (float)(lowerBound +
|
||||
((float)rand() / (float)RAND_MAX) *
|
||||
(upperBound - lowerBound));
|
||||
++data;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void generatePatternData(float *data, const int dimx, const int dimy,
|
||||
const int dimz, const float lowerBound,
|
||||
const float upperBound) {
|
||||
for (int iz = 0; iz < dimz; iz++) {
|
||||
for (int iy = 0; iy < dimy; iy++) {
|
||||
for (int ix = 0; ix < dimx; ix++) {
|
||||
*data = (float)(lowerBound +
|
||||
((float)iz / (float)dimz) * (upperBound - lowerBound));
|
||||
++data;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
bool fdtdReference(float *output, const float *input, const float *coeff,
|
||||
const int dimx, const int dimy, const int dimz,
|
||||
const int radius, const int timesteps) {
|
||||
const int outerDimx = dimx + 2 * radius;
|
||||
const int outerDimy = dimy + 2 * radius;
|
||||
const int outerDimz = dimz + 2 * radius;
|
||||
const size_t volumeSize = outerDimx * outerDimy * outerDimz;
|
||||
const int stride_y = outerDimx;
|
||||
const int stride_z = stride_y * outerDimy;
|
||||
float *intermediate = 0;
|
||||
const float *bufsrc = 0;
|
||||
float *bufdst = 0;
|
||||
float *bufdstnext = 0;
|
||||
|
||||
// Allocate temporary buffer
|
||||
printf(" calloc intermediate\n");
|
||||
intermediate = (float *)calloc(volumeSize, sizeof(float));
|
||||
|
||||
// Decide which buffer to use first (result should end up in output)
|
||||
if ((timesteps % 2) == 0) {
|
||||
bufsrc = input;
|
||||
bufdst = intermediate;
|
||||
bufdstnext = output;
|
||||
} else {
|
||||
bufsrc = input;
|
||||
bufdst = output;
|
||||
bufdstnext = intermediate;
|
||||
}
|
||||
|
||||
// Run the FDTD (naive method)
|
||||
printf(" Host FDTD loop\n");
|
||||
|
||||
for (int it = 0; it < timesteps; it++) {
|
||||
printf("\tt = %d\n", it);
|
||||
const float *src = bufsrc;
|
||||
float *dst = bufdst;
|
||||
|
||||
for (int iz = -radius; iz < dimz + radius; iz++) {
|
||||
for (int iy = -radius; iy < dimy + radius; iy++) {
|
||||
for (int ix = -radius; ix < dimx + radius; ix++) {
|
||||
if (ix >= 0 && ix < dimx && iy >= 0 && iy < dimy && iz >= 0 &&
|
||||
iz < dimz) {
|
||||
float value = (*src) * coeff[0];
|
||||
|
||||
for (int ir = 1; ir <= radius; ir++) {
|
||||
value += coeff[ir] * (*(src + ir) + *(src - ir)); // horizontal
|
||||
value += coeff[ir] * (*(src + ir * stride_y) +
|
||||
*(src - ir * stride_y)); // vertical
|
||||
value +=
|
||||
coeff[ir] * (*(src + ir * stride_z) +
|
||||
*(src - ir * stride_z)); // in front & behind
|
||||
}
|
||||
|
||||
*dst = value;
|
||||
} else {
|
||||
*dst = *src;
|
||||
}
|
||||
|
||||
++dst;
|
||||
++src;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Rotate buffers
|
||||
float *tmp = bufdst;
|
||||
bufdst = bufdstnext;
|
||||
bufdstnext = tmp;
|
||||
bufsrc = (const float *)tmp;
|
||||
}
|
||||
|
||||
printf("\n");
|
||||
|
||||
if (intermediate) free(intermediate);
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
bool compareData(const float *output, const float *reference, const int dimx,
|
||||
const int dimy, const int dimz, const int radius,
|
||||
const float tolerance) {
|
||||
for (int iz = -radius; iz < dimz + radius; iz++) {
|
||||
for (int iy = -radius; iy < dimy + radius; iy++) {
|
||||
for (int ix = -radius; ix < dimx + radius; ix++) {
|
||||
if (ix >= 0 && ix < dimx && iy >= 0 && iy < dimy && iz >= 0 &&
|
||||
iz < dimz) {
|
||||
// Determine the absolute difference
|
||||
float difference = fabs(*reference - *output);
|
||||
float error;
|
||||
|
||||
// Determine the relative error
|
||||
if (*reference != 0)
|
||||
error = difference / *reference;
|
||||
else
|
||||
error = difference;
|
||||
|
||||
// Check the error is within the tolerance
|
||||
if (error > tolerance) {
|
||||
printf("Data error at point (%d,%d,%d)\t%f instead of %f\n", ix, iy,
|
||||
iz, *output, *reference);
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
++output;
|
||||
++reference;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
Reference in New Issue
Block a user