i fount unreal engine code here:

http://www.cabalchronicle.com/download/InfinityBlade_UE3_WIN&IOS_SRC_By_MeGaMaX/InfinityBlade/Development/Src/D3D9Drv/Src/D3D9Viewport.cpp

void FD3D9DynamicRHI::EndDrawingViewport(FViewportRHIParamRef ViewportRHI,UBOOL bPresent,UBOOL bLockToVsync)
{
	DYNAMIC_CAST_D3D9RESOURCE(Viewport,Viewport);

	SCOPE_CYCLE_COUNTER(STAT_D3D9PresentTime);

	GPUFrameTiming.EndTiming();

	extern DWORD GGPUFrameTime;
	if ( GPUFrameTiming.IsSupported() )
	{
		QWORD GPUTiming = GPUFrameTiming.GetTiming();
		QWORD GPUFreq = GPUFrameTiming.GetTimingFrequency();
		GGPUFrameTime = appTrunc( DOUBLE(GPUTiming) / DOUBLE(GPUFreq) / GSecondsPerCycle );
	}
	else
	{
		GGPUFrameTime = 0;
	}

	check(DrawingViewport.GetReference() == Viewport);
	DrawingViewport = NULL;

	// Clear references the device might have to resources.
	Direct3DDevice->SetRenderTarget(0,*BackBuffer);
	Direct3DDevice->SetDepthStencilSurface(NULL);

	UnsetPSTextures();
	UnsetVSTextures();

	Direct3DDevice->SetVertexShader(NULL);

	ResetVertexStreams();

	Direct3DDevice->SetIndices(NULL);
	Direct3DDevice->SetPixelShader(NULL);

	#if WITH_PANORAMA
		extern void appPanoramaRenderHookRender(void);
		// Allow G4WLive to render the Live Guide as needed (or toasts)
		appPanoramaRenderHookRender();
	#endif

	// Tell D3D we're done rendering.
	Direct3DDevice->EndScene();

	if(bPresent)
	{
		// Present the back buffer to the viewport window.
		HRESULT Result = S_OK;
		if(Viewport->IsFullscreen())
		{
			Result = Direct3DDevice->Present(NULL,NULL,NULL,NULL);
		}
		else
		{
			RECT DestRect;
			if(GetClientRect((HWND)Viewport->GetWindowHandle(),&DestRect))
			{		
				RECT SourceRect;
				SourceRect.left		= SourceRect.top = 0;
				SourceRect.right	= Viewport->GetSizeX();
				SourceRect.bottom	= Viewport->GetSizeY();

				// Only present to the viewport if its client area isn't zero-sized.
				if(DestRect.right > 0 && DestRect.bottom > 0)
				{
					Result = Direct3DDevice->Present(&SourceRect,NULL,(HWND)Viewport->GetWindowHandle(),NULL);
				}
			}
		}

		// Detect a lost device.
		if(Result == D3DERR_DEVICELOST || Result == E_FAIL)
		{
			// This variable is checked periodically by the main thread.
			bDeviceLost = TRUE;
		}
		else
		{
			VERIFYD3D9RESULT(Result);
		}
	}

	// Wait for the GPU to finish rendering the previous frame before finishing this frame.
	FrameSyncEvent.WaitForCompletion();
	FrameSyncEvent.IssueEvent();

	// If the input latency timer has been triggered, block until the GPU is completely
	// finished displaying this frame and calculate the delta time.
	if ( GInputLatencyTimer.RenderThreadTrigger )
	{
		FrameSyncEvent.WaitForCompletion();
		DWORD EndTime = appCycles();
		GInputLatencyTimer.DeltaTime = EndTime - GInputLatencyTimer.StartTime;
		GInputLatencyTimer.RenderThreadTrigger = FALSE;
	}
}

what the effect of this code:

FrameSyncEvent.WaitForCompletion();
	FrameSyncEvent.IssueEvent();

if i comment them, the engine works well too.

so i want to know why they do this ?

It seems to be waiting for an evens that occurred at the beginning of the previous frame.

Whenever you send commands to the GPU, they are actually just being written into a command buffer. The GPU may be executing D3D/GL calls 1,2,3+ frames after the CPU makes those calls.

I would assume that this is there as a hack to ensure that the GPU is only ever 1 frame behind the CPU, and no more than that. It basically seems like a hack to disable tripple buffering or any other kind of excessive buffering hat a driver might automatically be performing. I guess this might help reduce input-to-screen latency on some systems?

void FD3D9EventQuery::IssueEvent()
{
	if( Query )
	{
		Query->Issue(D3DISSUE_END);
	}
}

void FD3D9EventQuery::WaitForCompletion()
{
	if( Query )
	{
		UBOOL bRenderingIsFinished = FALSE;
		while ( D3DRHI->GetQueryData(Query,&bRenderingIsFinished,sizeof(bRenderingIsFinished),TRUE) && !bRenderingIsFinished )
		{
		}
	}
}