Installing LinuxCNC On HP T610 Thin Client

its fine with Axis.  It's actually  fine with qtpyvcp. The "killer" is VTK which is a bit of a cpu hog. But I'm going to do some more tests of that particular component now I have working monokrom-plasma sim.

I don't mind putting some work in to get a decent lathe setup - but I'm a machinist 1st and computer boffin 20th! 

My current lathe is actually running a parallel port setup with pathpilot. I have to say having used pathpilot for a few years it is perfect for what I need and the conversational programming side of it is the best I have used. I'm now building a new lathe which will have some extra features, external controls, toolchanger etc. Unfortunately so much of the path pilot GUI has been custom coded in Python it is hard to find your way through it and even the simplest mods can take some work. Probe-basic-lathe looks good - but you need a high screen resolution/17" monitor to make it workable and I want to use a 15" touchscreen (like on pathpilot) which will be built into a panel with a small PC in the back.

Hence I was going to have a go at building my own screens in QTPYVCP and then use the Pathpilot python conversational code reworked as a basis for conversational programming. That might be a big dream but at the moment looks to be the only way to get what I want.

Joco wrote:

Scripts that I used/adjusted for this test.

This includes the hog cpu script and the linuxcnc wrapper script that needs to be used. This wrapper starts linuxcnc on core 2 then moves the UI task to Core 1. This needed to ensure a balance of UI responsiveness while the RT tasks own Core 2.

Cpu hog on core 2 ensures clock speeds are maxed while linuxcnc is running. A quirk of this architecture seems to be that maxing core 2 also maxes core 1. But the opposite is not true. An alternative setup is to use idle=poll on the kernwl boot. In which case cou hog is not needed. However the cpu will run hot. I was seeing 70C which is deemed “hot” by this cpu with 100C as critical.
The advantage of cpu hog is if you stop lcnc the wrapper kills hog and the cpu will start to cool.

You will also see from the h/w test file that kernel 4.19 was used. I also tests with 4.9.0-13 and noticed that clock speeds seemed to be maxed irrespective of using idle=poll on kernel boot or not.

Cheers - J
 

J
I have run several tests using graphical and text mode latency test. I am pretty sure, they are different. I just had a short glimpse on your scripts. I feel, when getting time for it, I might get a punch on improving my attempt to propose a - kinda - standardized testing environment for hardware and latency. Would you mind me talking part of your scripts into my effort?
Or even better: Join efforts?
Greez
chris

This is originally a tommylight script from:
forum.linuxcnc.org/18-computer/39370-scr...-of-computer-latency

So if you are looking for the latest and want to work improvements etc suggest you work there.

The other scripts I had in the mix are offered as is where is.

You want to use/improve them, knock your socks off.  

 

Joco wrote:

This is originally a tommylight script from:
forum.linuxcnc.org/18-computer/39370-scr...-of-computer-latency

 

That is his script, not mine !

Ah.  Wasn't sure from the way the start of the thread was written. So essentially .. have at it!