merged
This commit is contained in:
@@ -40,6 +40,7 @@
|
||||
|
||||
#include <stdio.h>
|
||||
#include <algorithm>
|
||||
#include <string.h>
|
||||
#include <math.h>
|
||||
#include "../timing.h"
|
||||
#include "stencil_ispc.h"
|
||||
@@ -66,9 +67,25 @@ void InitData(int Nx, int Ny, int Nz, float *A[2], float *vsq) {
|
||||
}
|
||||
|
||||
|
||||
int main() {
|
||||
int main(int argc, char *argv[]) {
|
||||
static unsigned int test_iterations[] = {3, 3, 3};//the last two numbers must be equal here
|
||||
int Nx = 256, Ny = 256, Nz = 256;
|
||||
int width = 4;
|
||||
|
||||
if (argc > 1) {
|
||||
if (strncmp(argv[1], "--scale=", 8) == 0) {
|
||||
float scale = atof(argv[1] + 8);
|
||||
Nx *= scale;
|
||||
Ny *= scale;
|
||||
Nz *= scale;
|
||||
}
|
||||
}
|
||||
if ((argc == 4) || (argc == 5)) {
|
||||
for (int i = 0; i < 3; i++) {
|
||||
test_iterations[i] = atoi(argv[argc - 3 + i]);
|
||||
}
|
||||
}
|
||||
|
||||
float *Aserial[2], *Aispc[2];
|
||||
Aserial[0] = new float [Nx * Ny * Nz];
|
||||
Aserial[1] = new float [Nx * Ny * Nz];
|
||||
@@ -79,24 +96,22 @@ int main() {
|
||||
float coeff[4] = { 0.5, -.25, .125, -.0625 };
|
||||
|
||||
InitData(Nx, Ny, Nz, Aispc, vsq);
|
||||
|
||||
//
|
||||
// Compute the image using the ispc implementation on one core; report
|
||||
// the minimum time of three runs.
|
||||
//
|
||||
double minTimeISPC = 1e30;
|
||||
#if 0
|
||||
for (int i = 0; i < 3; ++i) {
|
||||
for (int i = 0; i < test_iterations[0]; ++i) {
|
||||
reset_and_start_timer();
|
||||
loop_stencil_ispc(0, 6, width, Nx - width, width, Ny - width,
|
||||
width, Nz - width, Nx, Ny, Nz, coeff, vsq,
|
||||
Aispc[0], Aispc[1]);
|
||||
double dt = get_elapsed_mcycles();
|
||||
printf("@time of ISPC run:\t\t\t[%.3f] million cycles\n", dt);
|
||||
minTimeISPC = std::min(minTimeISPC, dt);
|
||||
}
|
||||
|
||||
printf("[stencil ispc 1 core]:\t\t[%.3f] million cycles\n", minTimeISPC);
|
||||
#endif
|
||||
|
||||
InitData(Nx, Ny, Nz, Aispc, vsq);
|
||||
|
||||
@@ -105,12 +120,13 @@ int main() {
|
||||
// the minimum time of three runs.
|
||||
//
|
||||
double minTimeISPCTasks = 1e30;
|
||||
for (int i = 0; i < 3; ++i) {
|
||||
for (int i = 0; i < test_iterations[1]; ++i) {
|
||||
reset_and_start_timer();
|
||||
loop_stencil_ispc_tasks(0, 6, width, Nx - width, width, Ny - width,
|
||||
width, Nz - width, Nx, Ny, Nz, coeff, vsq,
|
||||
Aispc[0], Aispc[1]);
|
||||
double dt = get_elapsed_mcycles();
|
||||
printf("@time of ISPC + TASKS run:\t\t\t[%.3f] million cycles\n", dt);
|
||||
minTimeISPCTasks = std::min(minTimeISPCTasks, dt);
|
||||
}
|
||||
|
||||
@@ -123,12 +139,13 @@ int main() {
|
||||
// minimum time.
|
||||
//
|
||||
double minTimeSerial = 1e30;
|
||||
for (int i = 0; i < 3; ++i) {
|
||||
for (int i = 0; i < test_iterations[2]; ++i) {
|
||||
reset_and_start_timer();
|
||||
loop_stencil_serial(0, 6, width, Nx-width, width, Ny - width,
|
||||
width, Nz - width, Nx, Ny, Nz, coeff, vsq,
|
||||
Aserial[0], Aserial[1]);
|
||||
double dt = get_elapsed_mcycles();
|
||||
printf("@time of serial run:\t\t\t[%.3f] million cycles\n", dt);
|
||||
minTimeSerial = std::min(minTimeSerial, dt);
|
||||
}
|
||||
|
||||
|
||||
@@ -1,181 +1,16 @@
|
||||
<?xml version="1.0" encoding="utf-8"?>
|
||||
<Project DefaultTargets="Build" ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<ItemGroup Label="ProjectConfigurations">
|
||||
<ProjectConfiguration Include="Debug|Win32">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>Win32</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Debug|x64">
|
||||
<Configuration>Debug</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|Win32">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>Win32</Platform>
|
||||
</ProjectConfiguration>
|
||||
<ProjectConfiguration Include="Release|x64">
|
||||
<Configuration>Release</Configuration>
|
||||
<Platform>x64</Platform>
|
||||
</ProjectConfiguration>
|
||||
</ItemGroup>
|
||||
<PropertyGroup Label="Globals">
|
||||
<ProjectGuid>{2ef070a1-f62f-4e6a-944b-88d140945c3c}</ProjectGuid>
|
||||
<Keyword>Win32Proj</Keyword>
|
||||
<RootNamespace>rt</RootNamespace>
|
||||
<ISPC_compiler Condition=" '$(ISPC_compiler)' == '' ">ispc</ISPC_compiler>
|
||||
<ISPC_file>stencil</ISPC_file>
|
||||
<default_targets>sse2,sse4-x2,avx1-i32x8</default_targets>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
<CharacterSet>Unicode</CharacterSet>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<UseDebugLibraries>true</UseDebugLibraries>
|
||||
<CharacterSet>Unicode</CharacterSet>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<UseDebugLibraries>false</UseDebugLibraries>
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
<CharacterSet>Unicode</CharacterSet>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
|
||||
<ConfigurationType>Application</ConfigurationType>
|
||||
<UseDebugLibraries>false</UseDebugLibraries>
|
||||
<WholeProgramOptimization>true</WholeProgramOptimization>
|
||||
<CharacterSet>Unicode</CharacterSet>
|
||||
</PropertyGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
|
||||
<ImportGroup Label="ExtensionSettings">
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
|
||||
<Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="PropertySheets">
|
||||
<Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
|
||||
<Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
|
||||
</ImportGroup>
|
||||
<ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="PropertySheets">
|
||||
<Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
|
||||
</ImportGroup>
|
||||
<PropertyGroup Label="UserMacros" />
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
|
||||
<LinkIncremental>true</LinkIncremental>
|
||||
<ExecutablePath>$(ProjectDir)..\..;$(ExecutablePath)</ExecutablePath>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
|
||||
<LinkIncremental>true</LinkIncremental>
|
||||
<ExecutablePath>$(ProjectDir)..\..;$(ExecutablePath)</ExecutablePath>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
|
||||
<LinkIncremental>false</LinkIncremental>
|
||||
<ExecutablePath>$(ProjectDir)..\..;$(ExecutablePath)</ExecutablePath>
|
||||
</PropertyGroup>
|
||||
<PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
|
||||
<LinkIncremental>false</LinkIncremental>
|
||||
<ExecutablePath>$(ProjectDir)..\..;$(ExecutablePath)</ExecutablePath>
|
||||
</PropertyGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
|
||||
<ClCompile>
|
||||
<PrecompiledHeader>
|
||||
</PrecompiledHeader>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>$(TargetDir)</AdditionalIncludeDirectories>
|
||||
<IntrinsicFunctions>true</IntrinsicFunctions>
|
||||
<FloatingPointModel>Fast</FloatingPointModel>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
</Link>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
|
||||
<ClCompile>
|
||||
<PrecompiledHeader>
|
||||
</PrecompiledHeader>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<Optimization>Disabled</Optimization>
|
||||
<PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>$(TargetDir)</AdditionalIncludeDirectories>
|
||||
<IntrinsicFunctions>true</IntrinsicFunctions>
|
||||
<FloatingPointModel>Fast</FloatingPointModel>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
</Link>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PrecompiledHeader>
|
||||
</PrecompiledHeader>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<FunctionLevelLinking>true</FunctionLevelLinking>
|
||||
<IntrinsicFunctions>true</IntrinsicFunctions>
|
||||
<PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>$(TargetDir)</AdditionalIncludeDirectories>
|
||||
<FloatingPointModel>Fast</FloatingPointModel>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<EnableCOMDATFolding>true</EnableCOMDATFolding>
|
||||
<OptimizeReferences>true</OptimizeReferences>
|
||||
</Link>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
|
||||
<ClCompile>
|
||||
<WarningLevel>Level3</WarningLevel>
|
||||
<PrecompiledHeader>
|
||||
</PrecompiledHeader>
|
||||
<Optimization>MaxSpeed</Optimization>
|
||||
<FunctionLevelLinking>true</FunctionLevelLinking>
|
||||
<IntrinsicFunctions>true</IntrinsicFunctions>
|
||||
<PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
|
||||
<AdditionalIncludeDirectories>$(TargetDir)</AdditionalIncludeDirectories>
|
||||
<FloatingPointModel>Fast</FloatingPointModel>
|
||||
</ClCompile>
|
||||
<Link>
|
||||
<SubSystem>Console</SubSystem>
|
||||
<GenerateDebugInformation>true</GenerateDebugInformation>
|
||||
<EnableCOMDATFolding>true</EnableCOMDATFolding>
|
||||
<OptimizeReferences>true</OptimizeReferences>
|
||||
</Link>
|
||||
</ItemDefinitionGroup>
|
||||
<ItemGroup>
|
||||
<CustomBuild Include="stencil.ispc">
|
||||
<FileType>Document</FileType>
|
||||
<Command Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
|
||||
$(ISPC_compiler) -O2 %(Filename).ispc -o $(TargetDir)%(Filename).obj -h $(TargetDir)%(Filename)_ispc.h --arch=x86 --target=sse2,sse4-x2,avx
|
||||
</Command>
|
||||
<Command Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
|
||||
$(ISPC_compiler) -O2 %(Filename).ispc -o $(TargetDir)%(Filename).obj -h $(TargetDir)%(Filename)_ispc.h --target=sse2,sse4-x2,avx
|
||||
</Command>
|
||||
<Outputs Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">$(TargetDir)%(Filename).obj;$(TargetDir)%(Filename)_sse2.obj;$(TargetDir)%(Filename)_sse4.obj;$(TargetDir)%(Filename)_avx.obj;$(TargetDir)%(Filename)_ispc.h</Outputs>
|
||||
<Outputs Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">$(TargetDir)%(Filename).obj;$(TargetDir)%(Filename)_sse2.obj;$(TargetDir)%(Filename)_sse4.obj;$(TargetDir)%(Filename)_avx.obj;$(TargetDir)%(Filename)_ispc.h</Outputs>
|
||||
<Command Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
|
||||
$(ISPC_compiler) -O2 %(Filename).ispc -o $(TargetDir)%(Filename).obj -h $(TargetDir)%(Filename)_ispc.h --arch=x86 --target=sse2,sse4-x2,avx
|
||||
</Command>
|
||||
<Command Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
|
||||
$(ISPC_compiler) -O2 %(Filename).ispc -o $(TargetDir)%(Filename).obj -h $(TargetDir)%(Filename)_ispc.h --target=sse2,sse4-x2,avx
|
||||
</Command>
|
||||
<Outputs Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">$(TargetDir)%(Filename).obj;$(TargetDir)%(Filename)_sse2.obj;$(TargetDir)%(Filename)_sse4.obj;$(TargetDir)%(Filename)_avx.obj;$(TargetDir)%(Filename)_ispc.h</Outputs>
|
||||
<Outputs Condition="'$(Configuration)|$(Platform)'=='Release|x64'">$(TargetDir)%(Filename).obj;$(TargetDir)%(Filename)_sse2.obj;$(TargetDir)%(Filename)_sse4.obj;$(TargetDir)%(Filename)_avx.obj;$(TargetDir)%(Filename)_ispc.h</Outputs>
|
||||
</CustomBuild>
|
||||
</ItemGroup>
|
||||
<Import Project="..\common.props" />
|
||||
<ItemGroup>
|
||||
<ClCompile Include="stencil.cpp" />
|
||||
<ClCompile Include="stencil_serial.cpp" />
|
||||
<ClCompile Include="../tasksys.cpp" />
|
||||
</ItemGroup>
|
||||
<Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
|
||||
<ImportGroup Label="ExtensionTargets">
|
||||
</ImportGroup>
|
||||
</Project>
|
||||
|
||||
Reference in New Issue
Block a user