No announcement yet.

Should serial port throughputs should be approx the same?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Should serial port throughputs should be approx the same?

    I'm running BIT on NT4.0 on a custom-made PC-board.
    COM2 & 4 throughput is over 1000 times more than COMs 5,6,7 throughput (when running tests simultaneously at the same BAUD).
    COM 1,2,3,4 came built in to the PC-card that is at the heart of the board. COMs 5,6,7,8 were added later by someone else.
    Does this indicate some kind of fault on COMs 5,6,7?

    I assume that for a given baud rate, the throughput should be approx the same for any port?

    FYI I am seeing:
    COM2,COM4 both have similar throughput (1800 or so Bytes/sec @ 115200)
    COM5 1000 times less (throughput of 1 to 4 Bytes/sec (@ 115200, during same simultaneous test session))
    FYI it also says DSR is always low, but that might just be because it hasn't been wired through.
    The CPU usage and memory usage as reported by task manager are fine.

  • #2
    The throughput is to a great extent determined by the settings in the duty cycle window in BurnInTest. So if you want to run the ports as hard as they can go, set the duty cycle to 100% for the serial port test. Then turn off all the other tests so as not to load the CPU with other tasks.

    If you are running at 100% duty cycle at 115200 baud then you should expect around 11000 bytes / sec.

    This level of data transfer is positively slow for a modern computer. Network connections, disk drives, USB, all run hundreds of times faster than a serial port. Using the serial port should put almost no load on your system and all serial ports should function at approximately the same speeds (for a particular baud rate).

    So yes. You seem to have a problem with your hardware. Try running COM5,6,7 by themselves with 100% duty cycle. You might also try swapping your loopback plugs around in case there is a faulty plug.



    • #3
      Thanks for your reply, you've confirmed what I thought & I'll forward the info on to our electronics guys who made the board, along with a reccomendation that they buy some BIT licenses (they're looking to produce 50 of these boards AFAIK and currently have no soak testing apps AFAIK). (I'd tried swapping the loopback plugs, and the problem doesn't move; I'm only running the serial port test; I'll get them to check the duty cycle is 100%, I think it was).