What gain is achieved by double buffering as opposed a single buffering?

Answer 1

Buffer is like bucket of data you write to at your own pace so that some other process can read from it at its own pace when it is needed. If you use a single buffer, you may end up overwriting what the other process did not read yet, or the other process may "catch up" with you before you finish writing your current set of data. Basically, you still need to carefully synchronize the two processes. With two buffers, you can write to one buffer while the other process is still reading the other buffer.

Answer 2

Double buffering simply means performing all writes to a temporary buffer, then overwriting the real buffer with your temporary buffer in a single operation. A typical scenario is in an owner-drawn view that requires complex drawing code. Normally, the background is erased and the drawing code redraws the entire image from scratch. This is known as single-buffering. While this fine under normal operation, whenever you resize your window you will trigger dozens of redraw events creating an ugly flicker in your view as it's erased and redrawn multiple times.

To eliminate this flicker you need to stop erasing the image and simply overwrite it with a new image which you create in a separate memory buffer, one that has the same device context as your view. the overwriting process is known as a "bit-blast". Double-buffer costs more memory, of course, but the end result is a view that does not flicker when resized, nor when an another window or object is dragged across your image. Note that although you are no longer erasing the image, the new image overwrites every pixel so there will be no artefacts left behind by the previous image so it does not matter. This is achieved by constructing a new buffer upon which to draw your new image so you always start with a "blank canvas".