Chapter: Embedded Systems Design : Buffering and other data structures

Double buffering

The problem with single buffering is that there is a tremen-dous overhead in managing the buffer in terms of maintaining pointers, checking the pointers against water marks and so on.

Double buffering

The problem with single buffering is that there is a tremendous overhead in managing the buffer in terms of maintaining pointers, checking the pointers against water marks and so on. It would be a lot easier to separate the filling from the extraction. It removes many of the contention checks that are needed and greatly simplifies the design. This is the idea behind double buffering.

Instead of a single buffer, two buffers are used with one allocated for filling and the second for extraction. The process works by filling the first buffer and passing a pointer to it to the extraction task or routine. This filled buffer is then simply used by the software to extract the data. While this is going on, the second buffer is filled so that when the first buffer is emptied, the second buffer will be full with the next set of data. This is then passed to the extraction software by passing the pointer. Many designs will recycle the first buffer by filling it while the second buffer is emptied. The process will add delay into the system which will depend on the time taken to fill the first buffer.

Care must be taken with the system to ensure that the buffer swap is performed correctly. In some cases, this can be done by passing the buffer pointer in the time period between filling the last entry and getting the next one. In others, water marks can be used to start the process earlier so that the extraction task may be passed to the second buffer pointer before it is completely filled. This allows it the option of accessing data in the buffer if needed instead of having to wait for the buffer to complete filling. This is useful when the extraction timing is not consistent and/or re-quires different amounts of data. Instead of making the buffers the size of the largest data structure, they can be smaller and the double buffering used to ensure that data can be supplied. In other words, the double buffering is used to give the appearance of the presence of a far bigger buffer than is really there.

Study Material, Lecturing Notes, Assignment, Reference, Wiki description explanation, brief detail

Embedded Systems Design : Buffering and other data structures : Double buffering |