You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Upon testing the maximum poll rate in practical local networks, I find that the limit is approx. 20 ms.
In this article (https://www.controlengeurope.com/article/139766/Speeding-up-Modbus-TCP.aspx), I read about the suggestion to open tcp-sockets using the TCP_NODELAY option. I succeed in setting this option on the slave side (which is programmed using Zephyr OS).
However, if the master side is polling, this side should allow acknowledge of every single package in order to speed up things.
After cloning the repository, it points out that the sockets are not created using the standard tcp library (I refer to linux). Neither socket() is used nor an #include <sys/socket.h> is found. So I guess that it's not that simple in mbpoll code to disable Nagle's algorithm.
Comments on this or alternative ideas are of course welcome.
The text was updated successfully, but these errors were encountered:
Upon testing the maximum poll rate in practical local networks, I find that the limit is approx. 20 ms.
In this article (https://www.controlengeurope.com/article/139766/Speeding-up-Modbus-TCP.aspx), I read about the suggestion to open tcp-sockets using the TCP_NODELAY option. I succeed in setting this option on the slave side (which is programmed using Zephyr OS).
However, if the master side is polling, this side should allow acknowledge of every single package in order to speed up things.
After cloning the repository, it points out that the sockets are not created using the standard tcp library (I refer to linux). Neither
socket()
is used nor an#include <sys/socket.h>
is found. So I guess that it's not that simple in mbpoll code to disable Nagle's algorithm.Comments on this or alternative ideas are of course welcome.
The text was updated successfully, but these errors were encountered: