direct x, getting fps

I have my render set up like this...

Code:

//-----------------------------------------------------------------------------
// Name: WinMain()
// Desc: The application's entry point
//-----------------------------------------------------------------------------
INT WINAPI WinMain( HINSTANCE hInst, HINSTANCE, LPSTR, INT )
{
    // Register the window class
    WNDCLASSEX wc = { sizeof(WNDCLASSEX), CS_CLASSDC, MsgProc, 0L, 0L,
                      GetModuleHandle(NULL), NULL, NULL, NULL, NULL,
                      "D3D", NULL };
    RegisterClassEx( &wc );

    // Create the application's window
    HWND hWnd = CreateWindow( "D3D", "D3D",
                              WS_OVERLAPPEDWINDOW, 150, 150, 500, 500,
                              GetDesktopWindow(), NULL, wc.hInstance, NULL );

    // Initialize Direct3D
    if( SUCCEEDED( InitD3D( hWnd ) ) )
    {
        //Program Functions
        ProgramStart();
        
        // Show the window
        ShowWindow( hWnd, SW_SHOWDEFAULT );
        UpdateWindow( hWnd );
        
        // Enter the message loop
        MSG msg;
        ZeroMemory( &msg, sizeof(msg) );
        while( msg.message!=WM_QUIT )
        {
            if( PeekMessage( &msg, NULL, 0U, 0U, PM_REMOVE ) )
            {
                TranslateMessage( &msg );
                DispatchMessage( &msg );
            }
            else
            {
                ProgramStep();  //Other stuff to happen before rendering
                Render();  //Actaully rendering of d3d objects
            }
        }
    }

    UnregisterClass( "D3D", wc.hInstance );
    return 0;
}

The stuff I need looked at is near the end of the code.

But I want to know if that is the correct place to put it and how I should put it there. It seems to be a bit laggy, and I don't know how to get the fps.

Any help of any sort is helpfull to me!

Thanks

seems like an ok spot to put everything. Getting your FPS can be pretty simple:
1) count the number of frames drawn so far
2) once you hit 1 second of elapsed time, display that frame count, and reset to 0

That's the quick and dirty, and will suffice unless you want to do some pretty deep algo testing.

But that is my problem, how do I count to 1 second without stoping everything else?

And I am glad to see I am on track so far.

Here is some pseudo-code to point you in the right direction:

Code:

DWORD time, lastTime, frames = 0;

// before main loop
// set time to current time
// set lastTime to current time

// Enter your main program loop

// inside your rendering loop
// set time to current time
// increment frames
// if (time - lastTime) > 1 second
//   lastTime = time
//   fps = frames
//   frames = 0

Look up QueryPerformanceCounter / QueryPerformanceFrequency for good timing functions.

Maybe I forgoten alot of c++ but QueryPerformanceCounter gives a LARGE_INTEGER a value. But when I try to subtract a LARGE_INTEGER it gives an error saying I'm trying to do a int type thing.

Not to good of a explainanation, but how do I use math operations on the LARGE_INTEGERS?

Here is one way to do it:

Code:

bool Display(float timeDelta)
{
  
  if (Device)
  {
    //Clear back buffer to BLACK R:0 G:0 B:0 or 0x00000000
    Device->Clear(0,0,D3DCLEAR_TARGET | D3DCLEAR_ZBUFFER,d3d::BLACK,1.0f,0);
   
	//Start scene
    
    Device->BeginScene();
    
    //Compute FPS
      ElapsedTime+=timeDelta;
      FrameCount++;
    
      if (ElapsedTime>=1.0f)
      {
        FPS=(float)FrameCount/ElapsedTime;
        ElapsedTime=0.0f;
        FrameCount=0;
      } 
      ....
      ....
}

ok I get how to do, but it is the matter of doing it now.

I will jiggle around with it and see what happens.

Originally Posted by Rune Hunter

Not to good of a explainanation, but how do I use math operations on the LARGE_INTEGERS?

You perform the operations on the QuadPart member of the LARGE_INTEGER structure if your compiler has built in support for 64-bit integers. Otherwise you'll have to operate on the LowPart and HighPart members which are the lower and upper 32 bits respectively.

When using QueryPerformanceCounter and QueryPerformanceFrequency and you are using MSVC you can (read should) use __int64 instead of LARGE_INTEGER. Just perform a cast to LARGE_INTEGER when sending it to the functions.

I'm using dev-c++. I will try using a type cast to int64 anyways.

EDIT: The type cast seems to have worked (no error) but now I need to know how to change a number into a LPSTR. It isn't alowing me to go like

(LPSTR)number

were number is the __int64 variable. Or if it is alowing me to do that (I get no error if I do) the variable is NULL, wich it shouldn't be cause I have it initialized as 0 and it is used.

A simple sprintf will do the trick or you can use any of the string utility classes or MFC string classes to convert from a numeric data type to a string type. Windows also has wsprintf and other string functions that will also do the trick.

My advice is to link in the MFC string class via the DLL option - which does not load all of MFC - just the portion that has the string class. The MFC string class is quite nice and very very easy to use.