User`s guide

Continuous Data Transfer in PCIS-DASK ? 41
The PCIS-DASK double buffer mode functions were designed according to the principle described above. If you use
AI_AsyncDblBufferMode/DI_AsyncDblBufferMode to enable double buffer mode, the following continuous AI/DI
function will perform double-buffered continuous AI/DI. You can call
AI_AsyncDblBufferHalfReady/DI_AsyncDblBufferHalfReady to check if data in the circular buffer is half full
and ready for copying to the transfer buffer. Then you can call
AI_AsyncDblBufferTransfer/DI_AsyncDblBufferTransfer to copy data from the ready half buffer to the
transfer buffer.
5.2.2 Single-Buffered Versus Double-Buffered Data Transfer
Single-buffered data transfer is the most common method for continuous data transfer. In single-buffered input
operations, a fixed number of samples are acquired at a specified rate and transferred into user’s buffer. After the user’s
buffer stores the data, the application can analyze, display, or store the data to the hard disk for later processing.
Single-buffered operations are relatively simple to implement and can usually take advantage of the full hardware speed
of the device. However, the major disadvantage of single-buffered operation is that the maximum amount of data that
can be input at any one time is limited to the amount of initially allocated memory allocated in driver and the amount of
free memory available in the computer.
In double-buffered operations, as mentioned above, the data buffer is configured as a circular buffer. Therefore, unlike
single-buffered operations, double-buffered operations reuse the same buffer and are able to input or output an infinite
number of data points without requiring an infinite amount of memory. However, there exits the undesired result of data
overwritten for double-buffered data transfer. The device might overwrite data before PCIS-DASK has copied it to the
transfer buffer. Another data overwritten problem occurs when an input device overwrites data that PCIS-DASK is
simultaneously copying to the transfer buffer. Therefore, the data must be processed by the application at least as fast
as the rate at which the device is reading data. For most of the applications, this requirement depends on the speed and
efficiency of the computer system and programming language.
Hence, double buffering might not be practical for high-speed input applications.