This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

[FAQ] PROCESSOR-SDK-DRA8X-TDA4X: Mapping algorithm to Jacinto 7 using OpenVX


I have two algorithms that I would like to use on a given core within J7. Should I wrap these algorithms into two different OpenVX nodes within my OpenVX graph or should they be a single node?

  • The answer will depend on the configuration of the algorithms within your OpenVX graph.

    1. In the case that the two C7x algorithms are executing sequentially, each of these algorithms can be wrapped inside a single OpenVX node or separated into multiple nodes. There are tradeoffs with each of these possibilities:
      1. Single node: the intermediate data objects can be allocated from a memory region with better locality than DDR. This can be accomplished by using the tivxMemAlloc API and specifying the memory region where this data will be allocated.
      2. Multiple nodes: this maintains greater flexibility in the event that these nodes will not always be used sequentially and in tandem in other graphs.
    2. If the case that the two C7x algorithms are not required to be executed serially, the two algorithms should be separated into individual nodes.