HRESULT __stdcall CStopwatchClassFactory::CreateInstance(IUnknown *pUnkOuter, REFIID riid, void **ppvObject)
{
HRESULT hr;
CStopwatch* pStopwatch = new CStopwatch;
hr = pStopwatch->QueryInterface(riid, ppvObject);
if (FAILED(hr))
{
delete pStopwatch;
}
return hr;
}
bool
CCondVarBase::wait(CStopwatch& timer, double timeout) const
{
// check timeout against timer
if (timeout >= 0.0) {
timeout -= timer.getTime();
if (timeout < 0.0)
return false;
}
return wait(timeout);
}
void BaseApp::Run()
{
CStopwatch timer;
int sum = 0;
int counter = 0;
int deltaTime = 0;
while (1)
{
timer.Start();
if (kbhit())
{
KeyPressed (getch());
if (!FlushConsoleInputBuffer(mConsoleIn))
cout<<"FlushConsoleInputBuffer failed with error "<<GetLastError();
}
UpdateF((float)deltaTime / 1000.0f);
Render();
Sleep(1);
while (1)
{
deltaTime = timer.Now();
if (deltaTime > 20)
break;
}
sum += deltaTime;
counter++;
if (sum >= 1000)
{
TCHAR szbuff[255];
StringCchPrintf(szbuff, 255, TEXT("FPS: %d"), counter);
SetConsoleTitle(szbuff);
counter = 0;
sum = 0;
}
}
}
void MeasureConcurrentOperation(
TCHAR* operationName, DWORD nThreads, OPERATIONFUNC operationFunc) {
HANDLE* phThreads = new HANDLE[nThreads];
SetThreadPriority(GetCurrentThread(), THREAD_PRIORITY_HIGHEST);
for (DWORD currentThread = 0; currentThread < nThreads; currentThread++) {
phThreads[currentThread] =
CreateThread(NULL, 0, ThreadIterationFunction, operationFunc, 0, NULL);
}
SetThreadPriority(GetCurrentThread(), THREAD_PRIORITY_NORMAL);
CStopwatch watch;
WaitForMultipleObjects(nThreads, phThreads, TRUE, INFINITE);
__int64 elapsedTime = watch.Now();
_tprintf(
TEXT("Threads=%u, Milliseconds=%u, Test=%s\n"),
nThreads, (DWORD)elapsedTime, operationName);
// Don't forget to clean up the thread handles
for (DWORD currentThread = 0; currentThread < nThreads; currentThread++) {
CloseHandle(phThreads[currentThread]);
}
delete phThreads;
}
bool GLoadToken::LoadBuffer(const TCHAR* strFileName)
{
CStopwatch stopwatch;
TCHAR Drive[MAX_PATH];
TCHAR Dir[MAX_PATH];
TCHAR FName[MAX_PATH];
TCHAR Ext[MAX_PATH];
_tsplitpath(strFileName, Drive, Dir, FName, Ext);
m_szDirName = Drive;
m_szDirName += Dir;
m_szName = FName;
m_szName += Ext;
LARGE_INTEGER FileSize;
m_hHandle = CreateFile(strFileName, GENERIC_READ, FILE_SHARE_READ, NULL, OPEN_EXISTING,
FILE_ATTRIBUTE_NORMAL, NULL);
if (m_hHandle != INVALID_HANDLE_VALUE)
{
GetFileSizeEx(m_hHandle, &FileSize);
UINT cBytes = FileSize.LowPart;
SAFE_NEW_ARRAY(m_pmbStaticMeshData, char, cBytes);
if (!m_pmbStaticMeshData)
{
DeleteBuffer();
return false;//E_OUTOFMEMORY;
}
DWORD dwBytesRead;
if (!ReadFile(m_hHandle, m_pmbStaticMeshData, cBytes, &dwBytesRead, NULL))
{
DeleteBuffer();
return false;
}
//SetFilePointer( m_hHandle, 0, 0, FILE_BEGIN );
}
else
{
return false;
}
TCHAR* szMeshData = NULL;
TCHAR* SearchString = NULL;
size_t convertedChars = 0;
SAFE_NEW_ARRAY(m_pwcStaticMeshData, TCHAR, FileSize.LowPart);
if (mbstowcs_s(&convertedChars, m_pwcStaticMeshData, FileSize.LowPart, m_pmbStaticMeshData, _TRUNCATE) == 0)
{
DeleteBuffer();
return false;
}
if (m_pwcStaticMeshData)
{
SAFE_DELETE_ARRAY(m_pmbStaticMeshData);
m_pwcTokenData = m_pwcStaticMeshData;
m_pwcSearchData = m_pwcStaticMeshData;
}
CloseHandle(m_hHandle);
m_hHandle = 0;
BeginToken();
stopwatch.Output(L"GLoadToken::LoadBuffer");
return true;
}
void CEngine::RunGameLoop()
{
// Enter the message loop
MSG msg;
ZeroMemory(&msg, sizeof(msg));
#ifdef USE_DISP_FPS
char text[256] = {"\0"};
#endif
while (msg.message != WM_QUIT)
{
if (PeekMessage(&msg, nullptr, 0U, 0U, PM_REMOVE))
{
TranslateMessage(&msg);
DispatchMessage(&msg);
}
else {
#ifdef USE_DISP_FPS
CStopwatch sw;
sw.Start();
#endif
GameTime().Update();
//キー入力を更新。
m_keyInput.Update();
m_skinModelDataResources.Update();
m_physicsWorld.Update();
CRenderContext& topRenderContext = m_renderContextArray[0];
CRenderContext& lastRenderContext = m_renderContextArray[m_numRenderContext - 1];
topRenderContext.SetRenderTarget(0, &m_mainRenderTarget[m_currentMainRenderTarget]);
//topRenderContext.SetRenderTarget(0, &m_backBufferRT);
topRenderContext.SetRenderTarget(1, NULL);
CGameObjectManager& goMgr = CGameObjectManager::Instance();
goMgr.Execute(
m_renderContextArray.get(),
m_numRenderContext,
m_renderContextMap.get(),
m_preRender,
m_postEffect
);
lastRenderContext.SetRenderTarget(0, &m_backBufferRT);
lastRenderContext.Clear(0, nullptr, D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER,
D3DCOLOR_RGBA(0, 0, 0, 0), 1.0f, 0
);
CopyMainRenderTargetToBackBuffer(lastRenderContext);
m_pD3DDevice->BeginScene();
//レンダリングコマンドのサブミット
for( int i = 0; i < m_numRenderContext; i++ ){
m_renderContextArray[i].SubmitCommandBuffer();
}
// 描画
#ifdef USE_DISP_FPS
m_fpsFont.Draw(text, 0, 0);
#endif
m_pD3DDevice->EndScene();
m_pD3DDevice->Present(nullptr, nullptr, nullptr, nullptr);
//
#ifdef USE_DISP_FPS
sw.Stop();
sprintf(text, "fps = %lf\n", 1.0f / sw.GetElapsed());
#endif
}
}
}
bool GGbsModel::Init(TCHAR* strFileName, TCHAR* strShaderName) {
HRESULT hr = S_OK;
//파싱, 필요한 처리 시작.
CStopwatch stopwatch;
//I_AseParser
I_GbsParser.InitGbsModel(strFileName, this);
//필요한 처리 끝.
stopwatch.Output(L"Init()");
// Create the sample state
D3D11_SAMPLER_DESC sampDesc;
ZeroMemory(&sampDesc, sizeof(sampDesc));
sampDesc.Filter = D3D11_FILTER_MIN_MAG_MIP_LINEAR;
sampDesc.AddressU = D3D11_TEXTURE_ADDRESS_WRAP;
sampDesc.AddressV = D3D11_TEXTURE_ADDRESS_WRAP;
sampDesc.AddressW = D3D11_TEXTURE_ADDRESS_WRAP;
sampDesc.ComparisonFunc = D3D11_COMPARISON_NEVER;
sampDesc.MinLOD = 0;
sampDesc.MaxLOD = D3D11_FLOAT32_MAX;
hr = g_pd3dDevice->CreateSamplerState(&sampDesc, m_pSamplerLinear.GetAddressOf());
if (FAILED(hr))
return hr;
// Compile the vertex shader
ID3DBlob* pVSBlob = NULL;
hr = CompileShaderFromFile(strShaderName, "VS", "vs_4_0", &pVSBlob);
if (FAILED(hr))
{
MessageBox(NULL,
L"The FX file cannot be compiled. Please run this executable from the directory that contains the FX file.", L"Error", MB_OK);
return hr;
}
// Create the vertex shader
hr = g_pd3dDevice->CreateVertexShader(pVSBlob->GetBufferPointer(), pVSBlob->GetBufferSize(), NULL, m_pVertexShader.GetAddressOf());
if (FAILED(hr))
{
pVSBlob->Release();
return hr;
}
// Define the input layout
D3D11_INPUT_ELEMENT_DESC layout[] =
{
{ "POSITION", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 0, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "NORMAL", 0, DXGI_FORMAT_R32G32B32_FLOAT, 0, 12, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "COLOR", 0, DXGI_FORMAT_R32G32B32A32_FLOAT, 0, 24, D3D11_INPUT_PER_VERTEX_DATA, 0 },
{ "TEXCOORD", 0, DXGI_FORMAT_R32G32_FLOAT, 0, 40, D3D11_INPUT_PER_VERTEX_DATA, 0 },
};
UINT numElements = ARRAYSIZE(layout);
// Create the input layout
hr = g_pd3dDevice->CreateInputLayout(layout, numElements, pVSBlob->GetBufferPointer(),
pVSBlob->GetBufferSize(), m_pVertexLayout.GetAddressOf());
pVSBlob->Release();
if (FAILED(hr))
return hr;
// Set the input layout
g_pImmediateContext->IASetInputLayout(m_pVertexLayout.Get());
// Compile the pixel shader
ID3DBlob* pPSBlob = NULL;
hr = CompileShaderFromFile(strShaderName, "PS", "ps_4_0", &pPSBlob);
if (FAILED(hr))
{
MessageBox(NULL,
L"The FX file cannot be compiled. Please run this executable from the directory that contains the FX file.", L"Error", MB_OK);
return hr;
}
// Create the pixel shader
hr = g_pd3dDevice->CreatePixelShader(pPSBlob->GetBufferPointer(), pPSBlob->GetBufferSize(), NULL, m_pPixelShader.GetAddressOf());
pPSBlob->Release();
if (FAILED(hr))
return hr;
if (m_vGeomObj.size() == 1)
SingleModelInit();
else
MultiModelInit();
return true;
};
void
CFileChunker::sendFileChunks(char* filename, IEventQueue* events, void* eventTarget)
{
std::fstream file(reinterpret_cast<char*>(filename), std::ios::in | std::ios::binary);
if (!file.is_open()) {
throw runtime_error("failed to open file");
}
// check file size
file.seekg (0, std::ios::end);
size_t size = (size_t)file.tellg();
// send first message (file size)
CString fileSize = intToString(size);
size_t sizeLength = fileSize.size();
CFileChunk* sizeMessage = new CFileChunk(sizeLength + 2);
char* chunkData = sizeMessage->m_chunk;
chunkData[0] = kFileStart;
memcpy(&chunkData[1], fileSize.c_str(), sizeLength);
chunkData[sizeLength + 1] = '\0';
events->addEvent(CEvent(events->forIScreen().fileChunkSending(), eventTarget, sizeMessage));
// send chunk messages with a fixed chunk size
size_t sentLength = 0;
size_t chunkSize = m_chunkSize;
CStopwatch stopwatch;
stopwatch.start();
file.seekg (0, std::ios::beg);
while (true) {
if (stopwatch.getTime() > PAUSE_TIME_HACK) {
// make sure we don't read too much from the mock data.
if (sentLength + chunkSize > size) {
chunkSize = size - sentLength;
}
// for fileChunk->m_chunk, the first byte is the chunk mark, last is \0
CFileChunk* fileChunk = new CFileChunk(chunkSize + 2);
char* chunkData = fileChunk->m_chunk;
chunkData[0] = kFileChunk;
file.read(&chunkData[1], chunkSize);
chunkData[chunkSize + 1] = '\0';
events->addEvent(CEvent(events->forIScreen().fileChunkSending(), eventTarget, fileChunk));
sentLength += chunkSize;
file.seekg (sentLength, std::ios::beg);
if (sentLength == size) {
break;
}
stopwatch.reset();
}
}
// send last message
CFileChunk* transferFinished = new CFileChunk(2);
chunkData = transferFinished->m_chunk;
chunkData[0] = kFileEnd;
chunkData[1] = '\0';
events->addEvent(CEvent(events->forIScreen().fileChunkSending(), eventTarget, transferFinished));
file.close();
}
int main(int argc, char *argv[])
{
int max_threads = -1;
int use_invasic = 1;
if (argc > 1)
use_invasic = atoi(argv[1]);
if (use_invasic)
std::cout << "INVASIC ACTIVATED" << std::endl;
else
std::cout << "INVASIC NOT ACTIVATED" << std::endl;
int verbose_level = 0;
if (argc > 2)
verbose_level = atoi(argv[2]);
/*
* initialize MPI
*/
int rank, size;
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
MPI_Comm_size(MPI_COMM_WORLD, &size);
if (size == 1)
{
std::cout << "run with `mpirun -n 2 ./build/client_mpi_tbb_release [use invasic (0/1)] [verbose level (-99 for fancy graphics)]`" << std::endl;
return -1;
}
CWorldScheduler_threaded *cWorldScheduler_threaded;
if (use_invasic > 0)
{
/*
* allocate & start iPMO server for 1st rank on numa-domain
*/
if (rank == 0)
{
if (use_invasic == 1)
{
cWorldScheduler_threaded = new CWorldScheduler_threaded;
std::cout << "RANK 0: starting worldscheduler" << std::endl;
cWorldScheduler_threaded->start(-1, verbose_level, true);
}
}
/*
* WAIT FOR WORLD SCHEDULER!
*/
MPI_Barrier(MPI_COMM_WORLD);
/*
* initialize iPMO client
*/
#if USE_OMP
cPmo = new CPMO_OMP(max_threads);
#else
cPmo = new CPMO_TBB(max_threads);
#endif
// initial setup request
cPmo->invade_blocking(1, 1024, 0, nullptr, (float)1);
}
CStopwatch cStopwatch;
cStopwatch.start();
/*
* start simulation
*/
for (int t = 0; t < 100; t++)
{
if (rank == 0)
std::cout << "Timestep: " << t << std::endl;
// compute some workload
int workload = (t*((rank+size-1)%size) + (100-t)*rank) % 100;
// for 4 cores only on my laptop, a linear imbalance is not working on dual-core systems
workload *= workload;
std::cout << " > rank " << rank << " workload: " << workload << std::endl;
if (use_invasic > 0)
{
// is some message about optimizations available?
cPmo->reinvade_nonblocking();
// request resource update with new workload
cPmo->invade_nonblocking(1, 1024, 0, nullptr, (float)workload);
}
int n;
if (use_invasic > 0)
{
n = cPmo->getNumberOfThreads();
}
else
{
n = sysconf(_SC_NPROCESSORS_ONLN)/size;
}
/*
* spread some bogus workload across n threads
*/
#if USE_OMP
#pragma omp parallel for schedule(static, 1)
for (int i = 0; i < n; i++)
{
CDummyWorkload::doSomeSqrt(991290391.0, workload/n);
}
#else
tbb::parallel_for(
0, n, 1,
[workload,n](int i)
{
CDummyWorkload::doSomeSqrt(991290391.0, workload/n);
}
);
#endif
std::cout << " > rank " << rank << " finished" << std::endl;
// do some stupid barrier
MPI_Barrier(MPI_COMM_WORLD);
}
double t = cStopwatch.getTimeSinceStart();
std::cout << "runtime: " << t << std::endl;
if (use_invasic > 0)
{
// NEVER MIX retreat() with nonblocking invade/reinvade!
// cPmo_TBB->retreat();
/*
* free iPMO client
*/
delete cPmo;
// wait for all clients to shutdown iPMO
MPI_Barrier(MPI_COMM_WORLD);
/*
* free iPMO server
*/
if (rank == 0)
{
if (use_invasic == 1)
delete cWorldScheduler_threaded;
}
}
MPI_Finalize();
return EXIT_SUCCESS;
}
bool
CXWindowsPrimaryScreen::showWindow()
{
assert(m_window != None);
CDisplayLock display(m_screen);
// raise and show the input window
XMapRaised(display, m_window);
// grab the mouse and keyboard. keep trying until we get them.
// if we can't grab one after grabbing the other then ungrab
// and wait before retrying. give up after s_timeout seconds.
static const double s_timeout = 1.0;
int result;
CStopwatch timer;
do {
// keyboard first
do {
result = XGrabKeyboard(display, m_window, True,
GrabModeAsync, GrabModeAsync, CurrentTime);
assert(result != GrabNotViewable);
if (result != GrabSuccess) {
LOG((CLOG_DEBUG2 "waiting to grab keyboard"));
ARCH->sleep(0.05);
if (timer.getTime() >= s_timeout) {
LOG((CLOG_DEBUG2 "grab keyboard timed out"));
XUnmapWindow(display, m_window);
return false;
}
}
} while (result != GrabSuccess);
LOG((CLOG_DEBUG2 "grabbed keyboard"));
// now the mouse
result = XGrabPointer(display, m_window, True, 0,
GrabModeAsync, GrabModeAsync,
m_window, None, CurrentTime);
assert(result != GrabNotViewable);
if (result != GrabSuccess) {
// back off to avoid grab deadlock
XUngrabKeyboard(display, CurrentTime);
LOG((CLOG_DEBUG2 "ungrabbed keyboard, waiting to grab pointer"));
ARCH->sleep(0.05);
if (timer.getTime() >= s_timeout) {
LOG((CLOG_DEBUG2 "grab pointer timed out"));
XUnmapWindow(display, m_window);
return false;
}
}
} while (result != GrabSuccess);
LOG((CLOG_DEBUG1 "grabbed pointer and keyboard"));
return true;
}
