pyrogue.streamConnect(dmaStream[lane][vc],dmaStream[lane][vc]) reduced performance

XMLWordPrintable

    • Type: Task
    • Resolution: resolved
    • Priority: Minor
    • Component/s: SW
    • None

      https://github.com/slaclab/pgp-pcie-apps/blob/master/software/scripts/PrbsTesting.py#L140

      If the software absorb the FW TX PRBS stream, I get 40 Gb/s (args.loopback=False)

      If I loopback the FW TX to FW RX, I get ~3.0Gb/s (args.loopback=True)

      I would expect this SW loopback mode to achieve ~40Gb/s as well.

      I confirmed in a "cat /proc/datadev_0" that the TX buffers are not starved

              Assignee:
              Ryan Herbst
              Reporter:
              Larry Ruckman
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: