ID3D11DeviceContext::Flush method (d3d11.h)

Sends queued-up commands in the command buffer to the graphics processing unit (GPU).

Syntax

void Flush();

Return value

None

Remarks

Most applications don't need to call this method. If an application calls this method when not necessary, it incurs a performance penalty. Each call to Flush incurs a significant amount of overhead.

When Microsoft Direct3D state-setting, present, or draw commands are called by an application, those commands are queued into an internal command buffer. Flush sends those commands to the GPU for processing. Typically, the Direct3D runtime sends these commands to the GPU automatically whenever the runtime determines that they need to be sent, such as when the command buffer is full or when an application maps a resource. Flush sends the commands manually.

We recommend that you use Flush when the CPU waits for an arbitrary amount of time (such as when you call the Sleep function).

Because Flush operates asynchronously, it can return either before or after the GPU finishes executing the queued graphics commands. However, the graphics commands eventually always complete. You can call the ID3D11Device::CreateQuery method with the D3D11_QUERY_EVENT value to create an event query; you can then use that event query in a call to the ID3D11DeviceContext::GetData method to determine when the GPU is finished processing the graphics commands.

Microsoft Direct3D 11 defers the destruction of objects. Therefore, an application can't rely upon objects immediately being destroyed. By calling Flush, you destroy any objects whose destruction was deferred. If an application requires synchronous destruction of an object, we recommend that the application release all its references, call ID3D11DeviceContext::ClearState, and then call Flush.

Deferred Destruction Issues with Flip Presentation Swap Chains

Direct3D 11 defers the destruction of objects like views and resources until it can efficiently destroy them. This deferred destruction can cause problems with flip presentation model swap chains. Flip presentation model swap chains have the DXGI_SWAP_EFFECT_FLIP_SEQUENTIAL flag set. When you create a flip presentation model swap chain, you can associate only one swap chain at a time with an HWND, IWindow, or composition surface. If an application attempts to destroy a flip presentation model swap chain and replace it with another swap chain, the original swap chain is not destroyed when the application immediately frees all of the original swap chain's references.

Most applications typically use the IDXGISwapChain::ResizeBuffers method for the majority of scenarios where they replace new swap chain buffers for old swap chain buffers. However, if an application must actually destroy an old swap chain and create a new swap chain, the application must force the destruction of all objects that the application freed. To force the destruction, call ID3D11DeviceContext::ClearState (or otherwise ensure no views are bound to pipeline state), and then call Flush on the immediate context. You must force destruction before you call IDXGIFactory2::CreateSwapChainForHwnd, IDXGIFactory2::CreateSwapChainForCoreWindow, or IDXGIFactory2::CreateSwapChainForComposition again to create a new swap chain.

Requirements

Requirement Value
Target Platform Windows
Header d3d11.h
Library D3D11.lib

See also

ID3D11DeviceContext