The information in this topic pertains only to applications written in C++.

Full-screen exclusive mode DirectDraw objects will take advantage of hardware acceleration regardless of whether they are running on the primary device or on a secondary device. However, they cannot use built-in support for spanning graphics operations across display devices. It is the application's responsibility to perform operations on the appropriate device.

When the normal cooperative level is set, DirectDraw uses hardware acceleration only when the window is wholly within the display area of the primary device. When a window straddles two or more monitors, all blits are done in emulation and performance can be significantly slower. This is necessarily the case, because hardware buffers cannot blit to a display surface controlled by another piece of hardware.

As long as you create the DirectDraw object for the null device—that is, pass NULL to DirectDrawCreate as the lpGUID parameter—DirectDraw will blit to the entire window regardless of where it is located. However, if the device is created by its actual GUID, this is not the case, and blit operations that cross an edge of the primary surface will be clipped (if you are using a clipper) or will fail, returning DDERR_INVALIDRECT.

Note  When you are blitting to a window in a MultiMon application, negative coordinates are valid when the logical location of the secondary monitor is to the left of the primary monitor.

To get the best performance in a windowed MultiMon application, you need to create a DirectDraw object for each device, maintain off-screen surfaces in parallel on each device, keep track of which part of the window resides on each device, and perform separate blits to each device.