Developer Guide for Intel® Data Analytics Acceleration Library 2016 Update 4

Algorithm Output

Output of an algorithm can be one or several models or numeric tables. To retrieve the results of the algorithm computation, call the getResult() method. To access specific results, use the get(Result ID) method with the appropriate Result ID identifier. In the distributed processing mode, to get access to partial results of the algorithm computation, call the getPartialResult() method on each computation node. For a full list of algorithm computation results available, refer to the description of an appropriate algorithm.

By default, all algorithms allocate required memory to store partial and final results. Follow these steps to provide user allocated memory for partial or final results to the algorithm:

  1. Create an object of an appropriate class for the results. For the classes supported, refer to the description of a specific algorithm.
  2. Provide a pointer to that object to the algorithm by calling the setPartialResult() or setResult() method as appropriate.
  3. Call the compute() method. After the call, the object created contains partial or final results.