This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Network Middleware TCP timing

Dear all,

We've switched from the previous network middleware to version 7.0.0beta.
In the new middleware two threads run (1x Core-Thread and 1x ETH-Thread)

Now I want to send out TCP messages at 100Hz.
The network threads seem to run at about 40Hz (25ms)
But I want them to run faster to service my 100Hz message-creation-task. Is this possible?
Or are there other ways of solving this issue?

Kind regards,

Gerrit

Parents
  • Contacted Keil support and got this reply:


    This bad performance using netTCP_Send() is probably due to a bug we found recently in our MW Network v7.0, which can be summarized as follows:
    These socket send functions do not wakeup the network core thread to immediately process the send request. Instead these send requests are processed at the next tick, which causes a bad send performance.

    Thought I post this in case other people stumbled upon the same performance issue.
    They provided me with a patched library.
    This patch will be included in v7.1 which will probably launch end of March 2016
    In the meantime contact Keil for the patched libraries.

    Kind regards,

    Gerrit

Reply
  • Contacted Keil support and got this reply:


    This bad performance using netTCP_Send() is probably due to a bug we found recently in our MW Network v7.0, which can be summarized as follows:
    These socket send functions do not wakeup the network core thread to immediately process the send request. Instead these send requests are processed at the next tick, which causes a bad send performance.

    Thought I post this in case other people stumbled upon the same performance issue.
    They provided me with a patched library.
    This patch will be included in v7.1 which will probably launch end of March 2016
    In the meantime contact Keil for the patched libraries.

    Kind regards,

    Gerrit

Children