lcnc2.9 latency-test bug ?
02 Jan 2020 17:22 #153799
by tuxcnc
lcnc2.9 latency-test bug ? was created by tuxcnc
Attachments:
Please Log in or Create an account to join the conversation.
- Todd Zuercher
- Away
- Platinum Member
Less
More
- Posts: 4962
- Thank you received: 1369
02 Jan 2020 18:15 - 02 Jan 2020 18:21 #153801
by Todd Zuercher
Replied by Todd Zuercher on topic lcnc2.9 latency-test bug ?
Are you sure you are running a realtime enabled kernal?
What result does the command "uname -a" show in a terminal?
What result does the command "uname -a" show in a terminal?
Last edit: 02 Jan 2020 18:21 by Todd Zuercher.
Please Log in or Create an account to join the conversation.
02 Jan 2020 19:21 - 02 Jan 2020 19:27 #153804
by tuxcnc
It is of course realtime and gives extremely low latency, not exceed 5000 on my hardware.
Picture I attached shows not impossible situation, because max_interval=base+max_jitter.
If jitter is 25000 and base is 25000 than max interval should be 50000.
So, at start, jitter measure must be bad, I suppose.
Maybe it is the same reason to "rtapi error" message at start of linuxcnc gui ?
This is a good measure, after "Reset Statistics" :
Replied by tuxcnc on topic lcnc2.9 latency-test bug ?
This is 4.14.148-rtai-amd64 patched from github.com/NTULINUX/RTAI/ and compiled myself.Are you sure you are running a realtime enabled kernal?
It is of course realtime and gives extremely low latency, not exceed 5000 on my hardware.
Picture I attached shows not impossible situation, because max_interval=base+max_jitter.
If jitter is 25000 and base is 25000 than max interval should be 50000.
So, at start, jitter measure must be bad, I suppose.
Maybe it is the same reason to "rtapi error" message at start of linuxcnc gui ?
This is a good measure, after "Reset Statistics" :
Attachments:
Last edit: 02 Jan 2020 19:27 by tuxcnc. Reason: add picture
Please Log in or Create an account to join the conversation.
- Todd Zuercher
- Away
- Platinum Member
Less
More
- Posts: 4962
- Thank you received: 1369
02 Jan 2020 20:00 #153809
by Todd Zuercher
Replied by Todd Zuercher on topic lcnc2.9 latency-test bug ?
I'm not so sure the Max Interval needs to necessarily be equal to sum of the Max Jitter and thread time, because the time intervals before and after the Max Jitter point may have been significantly shorter than planned.
Regardless, perhaps running the Latency-histogram or Latency-plot might be more telling (specifically the Latency-plot.)
Regardless, perhaps running the Latency-histogram or Latency-plot might be more telling (specifically the Latency-plot.)
Please Log in or Create an account to join the conversation.
13 Jan 2020 17:26 #154773
by tuxcnc
Replied by tuxcnc on topic lcnc2.9 latency-test bug ?
Attachments:
Please Log in or Create an account to join the conversation.
Time to create page: 0.328 seconds