Script for automated testing of computer latency.

More
5 years 8 months ago #199543 by tommylight
Replied by tommylight on topic Script for automated testing of computer latency.
Screen latency is a topic for gamers, not linuxcnc.
I feel a ban going on ... :)

Please Log in to join the conversation.

More
5 years 7 months ago #202120 by JohnnyCNC
Replied by JohnnyCNC on topic Script for automated testing of computer latency.
I acquired a HP Slimline S5000 computer with 4GB ram and decided to give it a try. I followed the Linuxcnc-on-linux-mint-19-3-quick-and-easy-installation. I installed a PCI gracphics card because the video would go nuts and the machine would freeze when the lcnc-hw script or I manually tried to make the glxgears full screen. When I run the script it looks like it runs OK and is getting decent latency but at the end of the script is shows an error. Any idea what if anything I can do about this?

Closing browser firefox with WID of 65011715
Done.
..IPDL protocol error: Handler returned error code!

###!!! [Parent][DispatchAsyncMessage] Error: PLayerTransaction::Msg_ReleaseLayer Processing error: message was deserialized, but the handler returned false (indicating failure)

IPDL protocol error: Handler returned error code!

###!!! [Parent][DispatchAsyncMessage] Error: PLayerTransaction::Msg_ReleaseLayer Processing error: message was deserialized, but the handler returned false (indicating failure)

IPDL protocol error: Handler returned error code!



Thanks
John

Please Log in to join the conversation.

More
5 years 7 months ago #202129 by seuchato
Replied by seuchato on topic Script for automated testing of computer latency.
John
I cross my fingers, stating, that soon my worries are over and I can finally revert to this topic for good.

Allow me a sideremark: According to this , the cpu looks like (ehem) old?

Now, LCNC is known to work on just plain ole hardware. Not sure yet, if it covers this type. However, something like "IPDL protocol error: " is unknown to me. I'll have to dig into this.
Kindly be patient. I NEED to get through with some obstacles in my life I can dedicate the time to this again.

Greez
Chris

Please Log in to join the conversation.

More
5 years 7 months ago #202130 by seuchato
Replied by seuchato on topic Script for automated testing of computer latency.
John
Kindly post the .hw file and say the last 50 lines of the .lat file.
Greez
chris

Please Log in to join the conversation.

More
5 years 7 months ago #202134 by seuchato
Replied by seuchato on topic Script for automated testing of computer latency.

Cadmanxu wrote: I have added the text from the .hw file to a text document. In the editor when making the answer it does not let me insert the file by the extension ...

Attachment not found

Attachment not found


Cadmanxu
Thanks for adding the .hw files the way you did. I take note: for easy attachment, the extension has to be ".txt" not ".hw" or ".lat".

"Unexpected realtime delay on task 0 with period 50000"
That tells you, either :
  • to increase the base time (try 75000, 100000) or abandon the idea to use the very PC for softstepping (ie. no parport based machine controlling).
  • work on your boot parameters: //GRUB_CMDLINE_LINUX_DEFAULT="quiet"// might need alterations
  • There is a ton of nicely documentet boot parameters, that you can and should try out. If, please report back, others might just need t see what you achieved. (Tia for this)

Greez
chris

Ps: why don't you disbale pcspkr and sound?

Please Log in to join the conversation.

More
5 years 7 months ago #202137 by JohnnyCNC
Replied by JohnnyCNC on topic Script for automated testing of computer latency.

seuchato wrote: John
Kindly post the .hw file and say the last 50 lines of the .lat file.
Greez
chris


seuchato I am in no hurry at all. I too am dealing with life issues. I lost my mother this week and I am just keeping myself busy. My office was discarding a lot of equipment as we move to a permanent work from home model. I got this PC and thought I would give it a try. Thank you for your work on this.

John

Attachment not found



Last 50 lines of .lat
Code:
566 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 567 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 568 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 569 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 570 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 570 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 571 secs servo min: -37.600 uS max: 33.500 uS sdev: 4.800 uS 572 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 573 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 573 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 574 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 575 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 576 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 577 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 577 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 578 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 579 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 580 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 581 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 581 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 582 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 583 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 584 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 585 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 585 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 586 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 587 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 588 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 588 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 589 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 590 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 591 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 592 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 592 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 593 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 594 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 595 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 595 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 596 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 597 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 598 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 599 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 599 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 600 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 601 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 602 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 603 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 603 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 604 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 605 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 606 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS

Please Log in to join the conversation.

More
5 years 7 months ago #202187 by seuchato
Replied by seuchato on topic Script for automated testing of computer latency.
JohnnyCNC

Life is sometimes really hard. I feel with you.

I see you got a "model name : AMD Athlon(tm) II X2 240 Processor". That's why you do not get any temperature readings - yet. Next version of the script will address that. My last Athlon stopped working around 2 years ago. So, I hope you could do some test runs for me when I am ready?

You obviuosly started playing around with the buttons available: "GRUB_CMDLINE_LINUX_DEFAULT="quiet isolcpus=1 splash"". I suggest, go off the beaten track. I'd even try "isolcpus=0" and other options.

Did you disable anything regarding hyperthreading, vtd, iommu, cpu sleep states, throttling etc. in BIOS? In bios, just disable *everything* that you really do not need. That usually does wonders.

Speaking of BIOS: Do you have other versions available? If, play with these. A word of caution though: on a hp 8300 sff it was not possible to downgrade a bios for "security" reasons.

All that stuff consumes a lot of time, I know. It may be worth it at the end of the day though.

Code:
605 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS 606 secs servo min: -37.600 uS max: 34.100 uS sdev: 4.800 uS
Hmm, strange that the .lat file does not protocol the real time delay.

greez
chris

Please Log in to join the conversation.

More
5 years 6 months ago #206344 by Joco
Replied by Joco on topic Script for automated testing of computer latency.
In the vein of automated testing and considering that one of the next key elements in a viable PC host for LinuxCNC when used with Ethernet boards such as 7i96 or 7i76e is the stability of the NIC. Does anyone have a gcode prog they have used to "thrash" the ethernet connection to a ethernet mesa card? This woud be done with no motors connected. Just to ensure your NIC has good stable conenction to the card. I have an HP T610 thin client that is looking pretty good but if the NIC side is not up to snuff then it’s dead in the water as no parallel port and no way to add one.

Once I have this final part of the puzzle solved/tested I will publish results.

Cheers - James.

Please Log in to join the conversation.

More
5 years 6 months ago #206345 by tommylight
Replied by tommylight on topic Script for automated testing of computer latency.
Usually i leave it on and running some gcode for two or more days non stop.
The HP 8300 SFF connected to my big plasma was on for over 2 months non stop, not running code most of the time but connected to Mesa 7i92 and never lost connection with 4.9 RT kernel. Same PC/machine with 5.10 kernel does loose the connection after a while, roughly 6 to 10 hours.

Please Log in to join the conversation.

More
5 years 6 months ago #206346 by Joco
Replied by Joco on topic Script for automated testing of computer latency.
Thanks Tommy. I'm a bit of a gcode newbie so I am assuming to run gcode for hours you have some truely massive code file or you have a relatively simple shape, say the standard test linucnc ngc file, that you have wrapped in what is effectively a for-next loop to run the same gcode 100s or 1000s of times?

Cheers - James.

Please Log in to join the conversation.

Time to create page: 0.144 seconds
Powered by Kunena Forum