LinuxCNC S-Curve Accelerations
- rodw
-
- Offline
- Platinum Member
-
Less
More
- Posts: 11306
- Thank you received: 3787
15 Apr 2025 18:07 #326392
by rodw
Replied by rodw on topic LinuxCNC S-Curve Accelerations
ANd also try installing this 6.12 kernel another user built
drive.google.com/drive/folders/16Xlry00L...SlwznPLv?usp=sharing
drive.google.com/drive/folders/16Xlry00L...SlwznPLv?usp=sharing
The following user(s) said Thank You: Grotius
Please Log in or Create an account to join the conversation.
- Grotius
-
- Offline
- Platinum Member
-
Less
More
- Posts: 2414
- Thank you received: 2331
15 Apr 2025 18:28 #326396
by Grotius
Replied by Grotius on topic LinuxCNC S-Curve Accelerations
Thanks,
Will give it a try tomorrow.
Will give it a try tomorrow.
Please Log in or Create an account to join the conversation.
- Grotius
-
- Offline
- Platinum Member
-
Less
More
- Posts: 2414
- Thank you received: 2331
16 Apr 2025 08:02 #326433
by Grotius
Replied by Grotius on topic LinuxCNC S-Curve Accelerations
Hi,
Before starting to change kernels etc. I thought maybe first do the latency-histogram test.
I did this on 2 different pc's.
From what i read here : max us > 100 = is bad.
Is this value both the same for base- & servo-thread?
Or maybe i am reading the histogram not correct.
Before starting to change kernels etc. I thought maybe first do the latency-histogram test.
I did this on 2 different pc's.
From what i read here : max us > 100 = is bad.
Is this value both the same for base- & servo-thread?
Or maybe i am reading the histogram not correct.
Attachments:
Please Log in or Create an account to join the conversation.
- tommylight
-
- Away
- Moderator
-
Less
More
- Posts: 20418
- Thank you received: 6940
16 Apr 2025 09:10 #326435
by tommylight
Replied by tommylight on topic LinuxCNC S-Curve Accelerations
Always use --show at the end of latency histogram, it will show the number of excursions that are not inside the visible area of the histogram, but they can be seen as multicolored columns on both sides of the histogram.
Also, use --nobase if there is no need for the base period, it adds more timing strains to CPU that will mess with the servo period latency results.
For both periodsFor servo period only
Also, use --nobase if there is no need for the base period, it adds more timing strains to CPU that will mess with the servo period latency results.
For both periods
latency-histogram --showlatency-histogram --nobase --sbinsize 1000 --show
The following user(s) said Thank You: akb1212, Grotius
Please Log in or Create an account to join the conversation.
- langdons
- Offline
- Platinum Member
-
Less
More
- Posts: 491
- Thank you received: 50
16 Apr 2025 14:02 - 16 Apr 2025 14:04 #326459
by langdons
Especially XTerm.
Maybe also UXTerm.
I use the terminal to set the cpu_scaling_governor to performance, which reduces latency a fair amount.
Replied by langdons on topic LinuxCNC S-Curve Accelerations
XTerm, SSH, and Samba could come in handy.No basically not for me ... I am using only Geany, wine and lcnc + ethercat-master but others... If I am understand right what do you mean ... other stuff should be installed then
Especially XTerm.
Maybe also UXTerm.
I use the terminal to set the cpu_scaling_governor to performance, which reduces latency a fair amount.
Last edit: 16 Apr 2025 14:04 by langdons. Reason: Added reason to include XTerm
Please Log in or Create an account to join the conversation.
- tommylight
-
- Away
- Moderator
-
Less
More
- Posts: 20418
- Thank you received: 6940
16 Apr 2025 14:54 #326464
by tommylight
Replied by tommylight on topic LinuxCNC S-Curve Accelerations
SSH and samba are security risks, hence usually not included in major distros.
They should be installed only by whoever actually needs them, and some security info should be digested before venturing into it.
They should be installed only by whoever actually needs them, and some security info should be digested before venturing into it.
Please Log in or Create an account to join the conversation.
- langdons
- Offline
- Platinum Member
-
Less
More
- Posts: 491
- Thank you received: 50
16 Apr 2025 16:18 - 16 Apr 2025 16:19 #326470
by langdons
Replied by langdons on topic LinuxCNC S-Curve Accelerations
Though "included" ≠ "enabled"
Last edit: 16 Apr 2025 16:19 by langdons.
Please Log in or Create an account to join the conversation.
- tommylight
-
- Away
- Moderator
-
Less
More
- Posts: 20418
- Thank you received: 6940
16 Apr 2025 16:29 #326474
by tommylight
Replied by tommylight on topic LinuxCNC S-Curve Accelerations
They are enabled when installed and fully functioning.Though "included" ≠ "enabled"
Please Log in or Create an account to join the conversation.
- snowgoer540
-
- Offline
- Moderator
-
Less
More
- Posts: 2452
- Thank you received: 815
17 Apr 2025 02:31 #326501
by snowgoer540
I’m just watching the sidelines from a car driving by the football game … but it seems to me like you’re doing an awful lot of work to avoid making a new git sign in and creating a pull request, etc.
It just seems like it further fractures and obfuscates an already fractured project that is kind of on life support as it is.
Just my $0.02 for what it’s worth. ¯\_(ツ)_/¯
Replied by snowgoer540 on topic LinuxCNC S-Curve Accelerations
@Rodw,
Thank you for your effort, all your input so far.
I have tried to create a .iso for day's, using the lb build system and also by using a chroot environment.
However it failed at some point when the .iso installation was ~60% complete.
Also the linuxcnc live build failed at some point, it had something todo with filmware, but that's no problem.
Then creating .deb packages is also a lot off work. As you have at least 2 types. The normal packages and the -dev packages
wich includes .headers etc.
I think we have to use 3 types, like the normal, the -dev and the -source.
For now i use a Qemu vm. Install debian and the software we need. Then the Qemu .img file can be compressed and
downloaded by others to test in their vm.
For running on real machines, i have to do a extra conversion step from .img to other .img type,
then write the .img to usb, and it will run.
I’m just watching the sidelines from a car driving by the football game … but it seems to me like you’re doing an awful lot of work to avoid making a new git sign in and creating a pull request, etc.
It just seems like it further fractures and obfuscates an already fractured project that is kind of on life support as it is.
Just my $0.02 for what it’s worth. ¯\_(ツ)_/¯
Please Log in or Create an account to join the conversation.
- Grotius
-
- Offline
- Platinum Member
-
Less
More
- Posts: 2414
- Thank you received: 2331
17 Apr 2025 08:02 #326510
by Grotius
Replied by Grotius on topic LinuxCNC S-Curve Accelerations
@SnowGoer540,
At the moment we have 2 testers who failed to run scurve on their machines. One off the testers
is running the cia402 with ethercat. Somehow linuxcnc has following errors using the tpmod_scurve.
At the moment we are unable to find the reason. Maybe kernel is too slow, mabye it's something else.
The other tester is anonymous and problably has multiple machines running. He pointed out the latency is very bad for
the latest debian 13 rt kernel. So i made a debian 10 and a 4.19 rt kernel wich he is familair with. But this also runs dramatic
on my laptop.
Then on my laptop i got also poor results using ethercat steppers at the moment.
In the past everything worked flawless, but this was on a normal desktop pc running ethercat and steppers.
Therefore maybe the problem is just the laptop, performing poor on the rt kernel.
Then there is the cmake build system wich i use to compile the scurve planner. The linuxcnc makefile is unable to
compile c++ code along with it. And i think the cmake build system for the scurve planner will not be merged to the linuxcnc
github source, because it doesn't align with the current makefile system.
Then there is that extra licence note, that will not be approved, but still has our attention.
So even when the scurve run's nice in a linuxcnc simulator config. There are still other problems that has to be resolved at the moment.
Therefore i think it's too early to do a git push request.
Personally i don't see linuxcnc as a fractured project. Best is too stay positive
At the moment we have 2 testers who failed to run scurve on their machines. One off the testers
is running the cia402 with ethercat. Somehow linuxcnc has following errors using the tpmod_scurve.
At the moment we are unable to find the reason. Maybe kernel is too slow, mabye it's something else.
The other tester is anonymous and problably has multiple machines running. He pointed out the latency is very bad for
the latest debian 13 rt kernel. So i made a debian 10 and a 4.19 rt kernel wich he is familair with. But this also runs dramatic
on my laptop.
Then on my laptop i got also poor results using ethercat steppers at the moment.
In the past everything worked flawless, but this was on a normal desktop pc running ethercat and steppers.
Therefore maybe the problem is just the laptop, performing poor on the rt kernel.
Then there is the cmake build system wich i use to compile the scurve planner. The linuxcnc makefile is unable to
compile c++ code along with it. And i think the cmake build system for the scurve planner will not be merged to the linuxcnc
github source, because it doesn't align with the current makefile system.
Then there is that extra licence note, that will not be approved, but still has our attention.
So even when the scurve run's nice in a linuxcnc simulator config. There are still other problems that has to be resolved at the moment.
Therefore i think it's too early to do a git push request.
Personally i don't see linuxcnc as a fractured project. Best is too stay positive
The following user(s) said Thank You: akb1212, Clive S, snowgoer540, endian
Please Log in or Create an account to join the conversation.
Time to create page: 0.194 seconds