[SOLVED]linuxcnc Latency test ambiguous results

NVIDIA proprietary drivers do not work with RT kernel and neither do new AMD drivers.
Most of the time NVIDIA cards will cause latency spikes and should not be used for machine control, but i have plenty of GTX1060 and GTX1070 that are usable with NOVEAU drivers, older ones will mess latency badly, and even the ones i have on some motherboards do cause issues, so it's a cr@p shoot.
First, fix the ethernet latency, disable power saving options in BIOS and hyperthreading and TPM and..., then we can see what still needs to be done.

installed [code]r8168:

ping -c 4 -i .2 10.10.10.10
PING 10.10.10.10 (10.10.10.10) 56(84) bytes of data.
64 bytes from 10.10.10.10: icmp_seq=1 ttl=255 time=0.298 ms
64 bytes from 10.10.10.10: icmp_seq=2 ttl=255 time=0.308 ms
64 bytes from 10.10.10.10: icmp_seq=3 ttl=255 time=0.273 ms
64 bytes from 10.10.10.10: icmp_seq=4 ttl=255 time=0.273 ms

--- 10.10.10.10 ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 610ms
rtt min/avg/max/mdev = 0.273/0.288/0.308/0.015 ms


chrt 99 ping -i .001 -q -c  60000 10.10.10.10
PING 10.10.10.10 (10.10.10.10) 56(84) bytes of data.

--- 10.10.10.10 ping statistics ---
60000 packets transmitted, 60000 received, 0% packet loss, time 59999ms
rtt min/avg/max/mdev = 0.178/0.253/0.441/0.031 ms

no packet loss this time with r8168!! thanks !

i have system with the nvidia proprietary driver i need to go back to nouveau?

DarkPhoinix wrote:

60000 packets transmitted, 60000 received, 0% packet loss, time 59999ms
rtt min/avg/max/mdev = 0.178/0.253/0.441/0.031 ms

Much better.

DarkPhoinix wrote:

i have system with the nvidia proprietary driver i need to go back to nouveau?

Most probably, but if it is working with RT kernel, do give it a try
latency-histogram --nobase --sbinsize 1000 --show
while playing youtube, skipping through it and maximizing/minimizing the video.

Thanks , this is result with nouveau(and rustdesk for remote desktop):
 

Looks perfectly fine, so do some tests with Mesa card attached and again do copy some stuff around, play video, etc, see if it drops the link.
If not, move on to making chips!

Junction error 1: This error only occurs when I watch YouTube videos. With only LinuxCNC running and CPU stress test (and rustdesk remote desktop), the job runs without errors until the end.
this is the end of my machine?

No, not yet.
Now is that point where you upload your configuration files here so we can have a look, joint errors are usually miss configuration on stepper systems.
What board are you using?

I still don't have the steppers or the drivers, I hope they will arrive soon, for now I have tested with only this stepper and driver connected:

 

thinking this was the problem I disconnected the stepper and left the card alone but the error is the same as soon as I start a video on YouTube or if I start chrt 99 ping -i .001 10.10.10.10 -c 40000000

the file conf:
 
 
 

driver:TB6560 JP-1635A
stepper: 57HD0401 15SB
controller: NVEM V5

Stepper and driver is only for test not final hardware.







 

Have you tried isolating the CPU cores?

I have the same processor, but a different motherboard, and the results are better.

It is possible that your problem is caused by something other than delays, but it is worth trying to improve the result.

 

I was just going down that road, I found a post where it was talked about, For now I did this: but I would like to try to add: rcu_nocbs

can you share your lspci result?