|Home > Introduction > PVBus C++ transaction and Tx_Result classes > TransactionGenerator efficiency considerations|
TransactionGenerators are most efficient for multiple accesses to one 4KB page.
Each TransactionGenerator caches connection information internally. This improves efficiency for multiple accesses to a single 4KB page. If a component requires repeated access data from different pages, for example when streaming from one location to another, ARM recommends you create a TransactionGenerator for each location.
You can dynamically create and destroy TransactionGenerators, but it is better to allocate them once at initialization and destroy them at shutdown. See the DMA example in the examples directory of the System Canvas Model Library distribution.