how to optimising long time servo-thread.tmax (mesa 7i92, i7 3770)

More
25 Sep 2024 03:02 #310742 by AnkerFly
Hello everyoneI'm concerned about the servo-thread.tmax value
I made a lot of adjustments. But I can’t count on getting good results with this equipment.
I collected information in the form of pictures.
I ask you to look and evaluate the result of my work.
Perhaps nothing needs to be done and the result is good.
Either I ask for help on how to achieve a better servo-thread.tmax
I apologize for my English.
 

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 03:04 #310744 by AnkerFly
I'll attach more pictures

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 03:06 #310745 by AnkerFly
and attach more pictures

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 05:08 - 25 Sep 2024 05:15 #310751 by PCW
Unless you get real time error or error finishing read errors when running LinuxCNC
those times should be fine.
 (worst total servo thread time (1236924 clocks) is about 350 usec which is fine for a 1 ms thread)
Last edit: 25 Sep 2024 05:15 by PCW.

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 06:15 #310755 by AnkerFly
Rarely in the past have I seen this error: Unexpected realtime delay on task 0 with period 1000000
However, after I configured the rt_setup.sh.txt script, the error stopped occurring.
The servo-thread.tmax time has been significantly reduced.
I bought an intel i7 3770. I was hoping to reduce the time.

However, I think there is no point in using i7. An intel i3 3240, which I used a month ago, will be enough. There seems to be no sense in the i7.

i3 3240 generates less heat

If you, as an expert, consider my result satisfactory, then I will stop there.

I have no one to turn to.
I just can't believe these are normal results.
After all, ping time is not bad.

Thanks for the answer.

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 06:21 - 25 Sep 2024 06:23 #310756 by AnkerFly
I can also add the following.
I noticed that when I run the program, the time is approximately 600,000.
If you start milling against the background of the graphics frame, where the trajectory is drawn, then the time increases to approximately 1,200,000You can see in the pictures what I have attached. There are several of them. With different names.
Last edit: 25 Sep 2024 06:23 by AnkerFly.

Please Log in or Create an account to join the conversation.

More
25 Sep 2024 13:57 #310773 by PCW
Yes, graphics often have a large effect on latency
If you plot the times with halscope you can see this
effect dynamically. The latency test typically does
no show these latency issues as it really only shows
dispatch latency, since it does no actual memory access
or I/O.

You can even see the the effect in LinuxCNCs motion controller

 

Please Log in or Create an account to join the conversation.

Time to create page: 0.065 seconds
Powered by Kunena Forum