Details

    • Sub-task
    • Status: Closed (View Workflow)
    • High
    • Resolution: Completed
    • None
    • None
    • None
    • None

    Description

      The VecOps are a first area where ROOT can exploit data parallelism, also for analysis.
      What is require to vectorise TVec operations in a fully transparent way for the user is:

      • Move template specialisations of TVec, operations involving those as well as math functions to an implementation file
      • Compile libVecOps with aggressive flags (Ofast)
      • Replace math functions with the VDT ones if VDT is available
      • Add the marchnative CMake flag to allow to compile ROOT with native flags since instruction sets such avx2 or fma capabilities can massively influence vectorisation performance

      Attachments

        Activity

          People

            amadio Guilherme Amadio
            dpiparo Danilo Piparo
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:
              Actual Start: