Programming with ALF

Data partitioning
An important part to solving data parallel problems using multiple accelerators is to figure out how to partition data across the accelerators. The ALF API does not automatically partition data, however, it does provide a framework so that you can systematically partition the data.
Accelerator buffer management
On the accelerator, the ALF accelerator runtime manages the data of the work blocks and the task for the compute kernel. You only need to focus on the organization of data and the actual computational kernel. The ALF accelerator runtime handles buffer management and data movement. However, it is still important that you have a good understanding of how each buffer is used and its relationship with the computational kernel.
When to use the overlapped I/O buffer
Using work blocks and order of function calls per task instance on the accelerator
Based on the characteristics of an application, you can use single-use work blocks or multi-use work blocks to efficiently implement data partitioning on the accelerators.
Modifying the work block parameter and context buffer when using multi-use work blocks
Double buffering on ALF
When transferring data in parallel with the computation, double buffering can reduce the time lost to data transfer by overlapping it with the computation time. The ALF runtime implementation on Cell BE architecture supports three different kinds of double buffering schemes.
Performance and debug trace