• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 99 subscribers
  • Views 740 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

tcpEcho thoughput

Richard Bland
Expert 2050 points

Hi,

I am evaluating TI RTOS NDK on a EK TM4C1294XL. Running tcpEcho and using my own test program on a windows PC I find I can only get about 1.2Mbits/second throughput. Faster than this and the tx buffers back up in the PC. My test program sends data over TCP every 10mS and I can vary the data size. A data size of about 1500 bytes is the max I can get to. (1500*8*100=1.2Mbits/sec).

I wonder has anyone else measured this?

I have tried increasing several different buffer and stack sizes, but nothing seems to make much difference.

It has to be said that 1500 bytes is close to the maximum TCP packet payload. I wonder if this has any bearing?

My project needs to be able to handle about 6Mbits/sec.

Thanks

Richard

  • 0
    TI__Genius 12865 points
    Hi Richard,

    Could you elaborate on your network topology - is the TM4C1294XL connected to the same router as your PC? Just wondering if you may be seeing delays introduced by multiple hops.

    Best regards,
    Vincent
  • 0 in reply to Vincent W.
    Expert 2050 points

    Hi Vincent,

    Thanks for your reply.

    My PC and the TM4C1294XL are coonected to the same 1G switch. But that's a good thought, I will try a test from PC to PC across the same switch.

    However, suspecting the RTOS overhead was the cause, I wrote an echo program using the LWIP stack "raw" TCP functions. Disappointingly this performed just about the same, ie it will only echo at about 1.2Mbits/sec. Faster than this the buffer backs up in the transmitting PC.

    I then removed the transmit side of the echo, simply throwing away the received packets. This showed the same problem, so it appears the slowness is in the receive code.

    I have also played with compile optimisations which also don't make much difference.

    Any ideas gratefully received!

    Richard

  • 0 in reply to Richard Bland
    Expert 2050 points
    Hi Vincent,
    Thanks for your suggestion. Testing PC to PC revealed the error was in the test software!
    I am now exceeding 6.4Mbits/sec through tcpEcho.
    Best regards
    Richard
  • 0 in reply to Richard Bland
    TI__Expert 7250 points
    There was a suggested answer and since there has been no active on this thread for more than a week, the suggested answer was marked as verify. Please feel free to select the "Reject Answer" button and reply with more details.