#linuxcnc-devel Logs
Jul 13 2019
#linuxcnc-devel Calendar
12:39 AM memleak: so andypugh built RTAI w/ kernel 4.14.132 + LinuxCNC debs
03:21 AM phillc54: linuxcnc-build: force build --branch=master 2000.docs
03:21 AM linuxcnc-build: build #4867 forced
03:21 AM linuxcnc-build: I'll give a shout when the build finishes
03:29 AM linuxcnc-build: Hey! build 2000.docs #4867 is complete: Warnings [8warnings compile]
03:29 AM linuxcnc-build: Build details are at http://buildbot.linuxcnc.org/buildbot/builders/2000.docs/builds/4867
01:35 PM Tom_itx is now known as Tom_L
04:45 PM memleak: andypugh, thank you for fixing that
04:45 PM andypugh: Thanks for spotting it.
04:46 PM memleak: in the next couple of days, I'll be adding kernel 4.19 support to my RTAI branch. To make it easy for you to pull those changes, it'd be a good idea to restart the fork on your end before then
04:47 PM memleak: I couldn't fix the new usage of cpu_isolated_map but Paolo fixed it in SVN.
04:48 PM memleak: Xenomai also pushed a 4.14 update with some more fixes and changes, and I'll be merging that in as well
04:48 PM memleak: so lots of good stuff coming up
04:51 PM memleak: when I rename the project, I'll make it under a new name and you'll have push rights to that for whatever debian fixes/changes you want to throw it
04:53 PM memleak: so it'll be like github.com/ARTI/ARTI-LVF -- Tempo-Reale Applicazione Interfaccia: La Versione Funzionante
04:53 PM memleak: *throw at it
04:54 PM memleak: *TRAI
04:54 PM memleak: ARTI was RTAI's name when it was still adeos (adeos real-time interface)
04:58 PM andypugh: Is there any situation where Xenomai is a better bet than RTAI? I seem to recall building a Xenomai kernel for the Udoo.
05:24 PM memleak: RTAI has no non-x86 support anymore
05:25 PM memleak: If you're using x86 there's no need for Xenomai
05:41 PM andypugh: Right, so Pi, Beaglebone, Udoo all need to use Preempt-RT or Xeonomai.
05:42 PM andypugh: I wonder to what extent thr RTAI folk are “finished” with RTAI? It’s an academic project, I think? Developed for a specific reason?
05:42 PM andypugh: I used to work in a department with it’s own dedicated Finite Element solver, with an in-house coder to look after it.
05:43 PM andypugh: But as I was leaving they were giving up on it, as the commercial offerings could do the same thing, more conveniently.
05:50 PM memleak: RTAI doesn't really need much anymore, it's still the lowest latency, FOSS solution for x86, supports 64-bit, the only possible issue in the future is with 16 or greater CPU cores, such as the Ryzen 3950X coming out later this year, which I hope to be able to afford at the end of the year.
05:50 PM memleak: RTAI is basically a finished project IMO, just needs to be updated to support the latest kernels
05:51 PM memleak: When IPIPE dies, then RTAI will die.
05:52 PM memleak: IPIPE won't be around for another 5 years, right? Xenomai will move toward PREEMPT_RT ?
06:02 PM cerna: Xenomai will not move only toward Preempt_rt.
06:16 PM cerna: andypugh: Have you tried if there is LinuxCNC latency difference (based on latency-test script) with and without X server running?
06:16 PM andypugh: No, not as such.
06:16 PM cerna: OK, thanks.
06:17 PM andypugh: Though, actually, I am just about to try running the latency test through SSH as my Pi4 has no screen
06:20 PM cerna: Yes, but then it will not be a difference, so no comparison. (Thing is, I am trying [non-RTAI] on i7-8550u NB and with X server I get around 50μs and without X server under 8μs.) So I am interested, if it is in RTAI the same.
06:25 PM memleak: All latency tests are affected by X, graphical apps, and switches VTs
06:26 PM memleak: A latency spike for example by opening up firefox is inevitable
06:26 PM memleak: Without X is always your best bet for latency, and don't hit control+alt+F{X}
06:29 PM cerna: Yeah, will have to try to isolate the task with cgroup and hope that the latency will be somewhere in the middle.
06:31 PM memleak: andypugh may not notice a spike with X because latency hits it's max on it's own without it, but that is hardware dependent.
06:32 PM memleak: X server itself has less of an impact than X apps
06:32 PM memleak: You may not notice any spike at all simply launching or starting X, but starting/stopping GDM, KDM or LightDM is another story
06:33 PM memleak: How would you go about launching Firefox in a cgroup?
06:34 PM memleak: One more thing, the reason why starting X with a DM vs. not a DM is because that VT switching happens when that DM gets started
06:34 PM memleak: you can run $ startx and not switch terminals because X starts on the VT you're currently on
06:36 PM cerna: I meant isolating the realtime task/thread to it's own CPU core. (Only thing is I have to read up on libcgroup and such.) What I have read so far, it should be creating new isolation group at system start and then adding firefox to it. Or not?
06:36 PM cerna: Booting into terminal and manually running startx is how I currently have my system set up.
06:38 PM andypugh: Hmm. I just installed the Buster .img from RealtimePi. It boots and all that, but LinuxCNC does not seem to think it is a realtime kernel.
06:38 PM andypugh: Linux realtimepi 4.19.50-v7l+ #895 SMP Thu Jun 20 16:03:42 BST 2019 armv7l GNU/Linux
06:39 PM cerna: andypugh: Can you read /sys/kernel/realtime ?
06:39 PM andypugh: No
06:40 PM cerna: Well, then it is not Preempt_RT. What about /dev/rtheap ?
06:41 PM cerna: For Xenomai.
06:41 PM andypugh: No
06:42 PM andypugh: Is there a Pi equivalent of grub?
06:45 PM cerna: Isn't Pi using u-boot? What do you want to do in grub?
06:46 PM andypugh: I was wondering if I need to _choose_ the RT kernel
06:47 PM cerna: Well, look into /boot/config.txt.
06:49 PM andypugh: Nothing there to do wth kernel choice.
06:49 PM andypugh: Just the things you see in raspi-config
06:50 PM cerna: Don't have Pi at hand, but adding kernel=name_of_kernel.img should boot into it.
06:50 PM memleak: U-Boot has boot params too
06:51 PM memleak: and depending on how you build the kernel, params can be built into the kernel as well, there's also an option in the kernel to not allow any arguments outside what is provided in the kernel
06:52 PM memleak: "realtimepi" is not very specific though.
06:53 PM memleak: if you're sure this is an IPIPE-enabled kernel (Xenomai) then there might be something in /proc
06:56 PM andypugh: I am not sure about _anything_
06:56 PM andypugh: Linuxcnc says
06:56 PM andypugh: using POSIX non-realtime
06:57 PM memleak: is there anything in /proc/ipipe ?
06:57 PM andypugh: I have a feeling that it might simply have not found the realtime during cinfugure
06:58 PM memleak: "realtimepi" is actually the host name and nothing to do with the kernel
06:58 PM memleak: it just says v7l which makes me think this is not an RT kernel
06:59 PM memleak: neither IPIPE nor PREEMPT_RT
07:00 PM cerna: It's this: https://github.com/guysoft/RealtimePi ? If so, they say Preempt_RT.
07:01 PM andypugh: Whan you say “in /proc/ipipe” would you expect it to show up all the time, using ls ?
07:02 PM memleak: not sure
07:09 PM andypugh: It is possible that I am an idiot
07:09 PM andypugh: I maye have neglected the “sudo make setuid”
07:11 PM andypugh: No, it wasn’t that. (which in no way excludes the possibility that I might be an idiot in other ways)
07:12 PM Tom_L: ok i've got stretch test pc up with preempt-rt on it. i can overwrite that with RTAI right?
07:13 PM andypugh: You can install both kernels and chose between them in grub
07:13 PM memleak: RTAI is a set of a kernel image and external (out-of-tree) modules
07:14 PM memleak: so what andypugh said
07:14 PM andypugh: sudo dpkg -i linux-image….deb
07:14 PM andypugh: and then the same for the rtai deb
07:14 PM memleak: andypugh, I noticed that the debs are based off your old fork
07:15 PM andypugh: No.
07:15 PM cerna: andypugh: Where did you get the Realtimepi image?
07:15 PM andypugh: They are stored in my old fork, but not built from it.
07:15 PM cerna: https://github.com/guysoft/RealtimePi/blob/53ea00ca903142add2049cef83e3a4a4de4f91e1/src/modules/realtimepi/config#L2 looks pretty RT.
07:16 PM andypugh: cerna: http://unofficialpi.org/Distros/RealtimePi/nightly/
07:16 PM memleak: Oh, they're just missing this commit then: https://github.com/NTULINUX/RTAI/commit/044ecf65a372d5fcd894a1027261ad8682e5117c
07:17 PM memleak: What is this docker stuff?
07:18 PM cerna: Never used? Not very sysadmin, are you?
07:18 PM memleak: Shouldn't he be using AWS with Kubernetes?
07:18 PM Tom_test: andypugh, so if i install linux-image-4.14.131-rtai-amd64_4.14.132-rtai-amd64-3_amd64.deb will that overwrite the current OS?
07:19 PM andypugh: Tom_test: No, it will just install an alternative kernel (and configure grub to boot from it by default)
07:20 PM memleak: cerna, I've been an admin for 12 years.
07:20 PM andypugh: It willl still boot into your current OS, just with a defferent kernel
07:20 PM Tom_test: did you move it?
07:20 PM andypugh: Hmm?
07:21 PM Tom_test: sudo dpkg --install linux-image-4.14.131-rtai-amd64_4.14.132-rtai-amd64-3_amd64.deb
07:21 PM Tom_test: error: cannot access archive
07:21 PM andypugh: Have you downloaded the deb?
07:22 PM Tom_test: i'm a rookie at this sry
07:22 PM cerna: memleak: No disrespect, just fun. It is just alpha and omega in admin departments in local companies. BTW, Kubernetes is something little different, I think.
07:22 PM andypugh: There are no repositories set up, you need to download the files and manually install.
07:22 PM Tom_test: ok
07:22 PM memleak: large companies, you mean.
07:23 PM memleak: the disorganized, heavily corporate ones.
07:23 PM andypugh: Google says that you can tell if the kernel has the RT patch by using uname… Even I know that is rubbish.
07:23 PM andypugh: It’s perfectly pssible to compile an RT kernel without RT in the name, and vice-versa
07:24 PM memleak: Well PREEMPT RT does show up if it's a PREEMPT RT kernel because it's hard-coded in
07:24 PM memleak: it's not part of the kernel string but the version
07:24 PM memleak: uname -v
07:24 PM cerna: Well, 2500 people in two local buildings. So, larger?
07:26 PM memleak: Docker came out 6 years ago, I wonder how Linux administrators had jobs before.
07:26 PM andypugh: memleak: Well, this realtime-pi image is showing “#895 SMP Thu Jun 20 16:03:42 BST 2019”
07:27 PM memleak: andypugh, that doesn't look like preempt_rt
07:27 PM andypugh: That’s what I am thinking, but then what is is doing in the realrtimepi repo?
07:28 PM memleak: I have no idea, I don't use other people's ARM scripts and tools.
07:29 PM Tom_test: andypugh, do i need to put the 'Assets' in a certain location?
07:29 PM memleak: Tom_test, dpkg -i can accept paths outside your current directory
07:29 PM memleak: the .debs can be installed from anywhere on your local system
07:31 PM andypugh: memleak: This is a fully assembled SD card image. (and ’tis true, it could be a terrible security risk, blindly installing it and connecting it to my network).
07:31 PM Tom_test: ahh, i think you updated the deb and not the sudo line
07:31 PM memleak: does the SD image have the kernel source installed?
07:31 PM memleak: conveniently, probably not. and no /proc/config.gz either
07:32 PM andypugh: memleak: I don’t know, and have closed it down now.
07:32 PM memleak: checking /boot might be of help too.
07:32 PM Tom_test: ok, off and running
07:33 PM Tom_test: i need the headers as well?
07:33 PM andypugh: in /boot there is kernel, kernel7 and kernel7l
07:34 PM andypugh: Tom_test: I don’t know. So try not…
07:35 PM andypugh: It’s not like I am a real programmer or computer admin. I am a physicist masquerading as an engineer. :-)
08:24 PM memleak: Tom_test, any success?
08:28 PM * cerna uploaded an image: modules.png (12KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/smWLomhdVevMbGxTNtdzyIcr >
08:28 PM cerna: It looks like the RTPI have several images.
08:28 PM cerna: Andypugh had Linux realtimepi 4.19.50-v7l+ #895 SMP Thu Jun 20 16:03:42 BST 2019 armv7l GNU/Linux
08:29 PM andypugh: Where did you find those?
08:31 PM cerna: It's tree of kernel modules.
08:31 PM andypugh: Where?
08:32 PM memleak: /lib/mod?
08:32 PM andypugh: Wierd
08:33 PM memleak: you can have more modules for different kernels than kernels available to the user to choose from as well
08:33 PM memleak: what matters is what's in /boot as well
08:34 PM andypugh: Yes, but, if I download an SD image for the Pi from RealtimePi I rather expected the system to boot up from a realtime kernel…
08:34 PM memleak: don't assume anything from anyone who posts things related to ARM on the internet
08:35 PM Tom_test: uname -a : Linux stretch 4.14.132-rtai-amd64 #1 SMP PREEMPT Mon Jul 8 23:47:38 BST 2019 x86_64 GNU/Linux
08:36 PM memleak: I trust Russell King but that's about it.
08:36 PM cerna: I ma trying to somehow open the .img kernel images, so far it is learning experience.
08:36 PM memleak: Tom_test, HOORAY!
08:36 PM andypugh: I only went there as Gene on the mailing list said that the results were better than a simply-patched standard kernel (one would guess through a good choice of kernel config stuff)
08:37 PM memleak: If I understand the docker stuff correctly, the only change he made is what's in the config export sym file
08:37 PM Tom_test: Starting LinuxCNC...
08:37 PM Tom_test: insmod: ERROR: could not insert module /usr/realtime-4.14.132-rtai-amd64/modules/rtai_hal.ko: Operation not permitted
08:38 PM memleak: https://raw.githubusercontent.com/guysoft/RealtimePi/devel/src/config
08:38 PM andypugh: Which debs have you installled this time?
08:38 PM memleak: wrong one..
08:38 PM memleak: https://raw.githubusercontent.com/guysoft/RealtimePi/devel/src/modules/realtimepi/config
08:38 PM Tom_test: i may need the headers
08:39 PM memleak: all he did was disable tracer which is off by default and specified the already default value of 21
08:39 PM andypugh: Hmm, so perhaps I am wastging my time with that repo then.
08:39 PM Tom_test: linux-image-4.14.132-rtai-amd64_4.14.132-rtai-amd64-1_amd64.deb
08:39 PM Tom_test: linuxcnc_2.8.0.pre1.4891.gadf1eff73_amd64.deb
08:39 PM andypugh: That said, Gene was taling abiut Stretch, not Buster.
08:40 PM Tom_test: rtai-modules-4.14.132_5.2.2-linuxcnc_amd64.deb
08:40 PM Tom_test: i think just those 3
08:40 PM andypugh: Tom_test: OK, try the headers then.
08:40 PM memleak: to insmod you need root/sudo btw
08:40 PM memleak: oh linuxcnc is doing it, nvm
08:40 PM memleak: you ran sudo make setuid ?
08:41 PM andypugh: He is installing debs, so can’t
08:41 PM memleak: sorry tired.. andypugh already did that
08:41 PM memleak: yes, that's in the debs, sorry
08:41 PM Tom_test: same errors
08:41 PM Tom_test: starting axis sim
08:41 PM andypugh: OK, so the debs work for me, but not for anyone else :-(
08:42 PM andypugh: Not really a surprise, but disappointing.
08:42 PM Tom_test: Starting LinuxCNC...
08:42 PM Tom_test: insmod: ERROR: could not insert module /usr/realtime-4.14.132-rtai-amd64/modules/rtai_hal.ko: Operation not permitted
08:42 PM Tom_test: Realtime system did not load
08:43 PM Tom_test: latency-test
08:43 PM Tom_test: insmod: ERROR: could not insert module /usr/realtime-4.14.132-rtai-amd64/modules/rtai_hal.ko: Operation not permitted
08:43 PM andypugh: Just for fun… sudo latency-test ?
08:43 PM Tom_test: same thing
08:44 PM andypugh: Can you reboot and try again?
08:46 PM memleak: I force enabled the rtai testsuite for this very reason :)
08:46 PM andypugh: Hmm, Google suggests that this could be because the modues are unsigned
08:47 PM memleak: lets have him try the rtai testsuite
08:47 PM Tom_test: ok
08:47 PM memleak: sudo /usr/realtime-4.14.132-rtai-amd64/testsuite/run
08:48 PM Tom_test: same numbers as before
08:48 PM memleak: operation not permitted? they dont need to be signed.
08:48 PM Tom_test: RTH| lat min| ovl min| lat avg| lat max| ovl max| overruns
08:48 PM Tom_test: RTD| 79| 79| 8388608| 64| 64| 0
08:48 PM memleak: is that the only line?
08:48 PM Tom_test: same as andy's i believe
08:48 PM Tom_test: no it scrolls continuously
08:49 PM memleak: and it stays at 64?
08:49 PM Tom_test: yes
08:49 PM memleak: wait.. are you using his no-calibration tree?
08:49 PM Tom_test: no not that i'm aware of
08:49 PM memleak: what does dmesg say?
08:49 PM Tom_test: i removed it and went with his others
08:49 PM memleak: ah ok
08:49 PM Tom_test: dmesg: read kernel buffer failed: Operation not permitted
08:50 PM memleak: sudo dmesg
08:51 PM memleak: SECURITY_DMESG_RESTRICT is enabled
08:51 PM memleak: andypugh, didn't you have good latency test scores?
08:52 PM andypugh: Yes.
08:52 PM andypugh: In the LinuxCNC latency test.
08:53 PM andypugh: The RTAI latency test looked spurious, as his does,
08:53 PM Tom_test: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/
08:57 PM andypugh: All I can find on Google is a suggestion that if Secure Boot is enabled in the BIOS then you can get this inserting nsigned modules.
08:59 PM andypugh: Time for me to be asleep, I am afraid.
08:59 PM andypugh: Sorry for wasting your time.
08:59 PM Tom_L: no waste here. we are learning
09:03 PM memleak: Tom_L, make sure secure boot is off in BIOS as well as legacy oprom only
09:03 PM memleak: no EFI stuff
09:04 PM Tom_test: lemme go pull it up...
09:05 PM memleak: [ 141.289221] RTAI[hal]: RTAI CONFIGURED WITH LESS THAN NUM ONLINE CPUS.
09:05 PM memleak: that's why
09:05 PM memleak: have to wait for tom to tell him that
09:05 PM memleak: ah he has hyperthreading on
09:06 PM Tom_L: how do i turn it off?
09:06 PM memleak: RTAI is configured for 4 core systems, he has 4 core CPU, but it has 8 threads
09:06 PM memleak: hey you're still here!
09:06 PM Tom_L: always
09:06 PM memleak: haha, in BIOS
09:06 PM Tom_L: it's up
09:06 PM Tom_L: (BIOS)
09:07 PM memleak: turn off hyperthreading and any EFI stuff including secure boot
09:07 PM memleak: flush keys, etc
09:07 PM memleak: and any power savings / eco friendly, cpu idle, c-state, power state, cpu frequency scaling, cpu boost, throttling etc
09:08 PM Tom_L: under advanced cpu core settings?
09:08 PM memleak: i don't know your BIOS, don't know what settings are available, but where ever they may be located, if any, if all, disable everything related to those as you can.
09:09 PM memleak: that would be a good guess for some of those options, yes.
09:09 PM Tom_L: ok hyperthreading disabled
09:09 PM memleak: already or just now?
09:09 PM Tom_L: just now
09:09 PM memleak: ok good
09:09 PM Tom_L: what else?
09:09 PM memleak: power savings, eco friendly, tree hugging environmentalist features
09:10 PM memleak: disable the "save the planet" setting :)
09:11 PM Tom_L: they obscure those from immediate view
09:11 PM Tom_L: turbo boost technology?
09:11 PM memleak: yeah the CPU should stay at one speed, in one state, all the time
09:12 PM memleak: no C6/C5 hibernate, suspend (if available)
09:13 PM Tom_L: i don't see anything else under advanced cpu settings
09:13 PM memleak: ok, then you should be all set!
09:13 PM Tom_L: c3 state support on
09:13 PM Tom_L: c6/7 state support on
09:13 PM Tom_L: auto
09:13 PM Tom_L: c8 auto
09:14 PM Tom_L: leave or change those?
09:14 PM Tom_L: all auto right now
09:14 PM memleak: all off
09:15 PM memleak: definitely all off
09:15 PM Tom_L: CPU enhanced hault
09:15 PM Tom_L: halt
09:16 PM Tom_L: (C1E)
09:16 PM Tom_L: auto
09:16 PM memleak: off
09:16 PM memleak: virtualization / sve / svm also off
09:17 PM memleak: (or IVT)
09:17 PM Tom_L: too late.. booting
09:19 PM Tom_test: latency test is running
09:19 PM memleak: i should add all those BIOS options to disable to the README file
09:19 PM memleak: and well?
09:20 PM memleak: RTAI or linuxcnc test?
09:20 PM Tom_test: lcnc
09:21 PM memleak: can you try the rtai test suite?
09:21 PM Tom_test: one sec
09:21 PM memleak: if it's broken for both of you guys, maybe i should just remove it.. it works here
09:22 PM Tom_test: what was the command line again?
09:22 PM memleak: sudo /usr/realtime-4.14.132-rtai-amd64/testsuite/run
09:22 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/latency_test.png
09:23 PM Tom_test: much different results
09:24 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/RTAI_Latency_test_suite.txt
09:25 PM memleak: they look close to me?
09:25 PM Tom_L: but you know what you're seeing :)
09:26 PM Tom_L: so fixing the bios helped alot
09:26 PM memleak: "ovl max" is equivalent to linuxcnc's "max jitter"
09:26 PM Tom_L: pretty decent i think
09:26 PM memleak: indeed!
09:26 PM Tom_L: no load though
09:26 PM memleak: so the test suite is acting weird for andypugh and i wish i knew why
09:27 PM Tom_L: has he turned things off in the bios?
09:27 PM memleak: and it sucks that he thought this was a waste of time but it really wasn't at all
09:27 PM Tom_L: well he'll wake up to a bright new day won't he?
09:27 PM memleak: heh yeah
09:30 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/Linuxcnc_RTAI_Stretch.png
09:30 PM memleak: so all is good!
09:30 PM Tom_L: it loads
09:31 PM Tom_L: numbers look good
09:31 PM Tom_L: the linuxcnc community thanks you all for all your efforts
09:32 PM memleak: i don't know the cnc side of lcnc at all, but how come when i load the 3dtest.ngc test sim, it says program exceeds machine maximum on axis z?
09:32 PM cerna: Hmm, "strings kernel7.img | grep 'Linux version 4.19.50-rt22-v7 (root@a96338bdce7f) (gcc version 8.3.0 (Raspbian 8.3.0-6+rpi1)) #2 SMP PREEMPT RT Sat Jun 29 10:17:23 BST 2019'
09:32 PM cerna: Linux version 4.19.50-rt22-v7 (root@a96338bdce7f) (gcc version 8.3.0 (Raspbian 8.3.0-6+rpi1)) #2 SMP PREEMPT RT Sat Jun 29 10:17:23 BST 2019", so it has preempt.
09:32 PM Tom_L: limits are likely not set
09:32 PM Tom_L: you should be able to disable that in the config i think
09:32 PM Tom_L: beyond that i'm not sure
09:33 PM Tom_test: let me try and load it
09:33 PM memleak: cerna: "#2 SMP PREEMPT RT"
09:34 PM memleak: andypugh was running #eight hundred and something non-preempt rt
09:34 PM Tom_test: memleak, where's it at?
09:35 PM memleak: nc_files/examples/3dtest.ngc
09:35 PM Tom_test: using the axis sim?
09:35 PM memleak: yes
09:36 PM Tom_test: i'm not sure, it loaded fine here
09:36 PM memleak: some of the others work fine as-is
09:37 PM memleak: you also probably have real configs set up
09:37 PM memleak: i'm using vanilla linuxcnc with no machine configs or anything, just launch, point, click
09:37 PM Tom_L: same here
09:38 PM Tom_L: i loaded the sim
09:38 PM Tom_test: well it's possible it sees my machine config
09:38 PM memleak: yeah thats what im guessing haha
09:38 PM Tom_test: lemme move em and try again
09:38 PM memleak: how long have you been using linuxcnc? i know i've seen you in this channel for years
09:39 PM Tom_test: i've followed it for years
09:39 PM Tom_test: started using it some time back
09:39 PM Tom_test: hard to say just how long
09:39 PM Tom_test: i recall when the govt had it
09:40 PM memleak: is machining for you a hobby or do you use it professionally?
09:40 PM Tom_test: i helped a friend with his cnc shop for a while, wouldn't say full time but i do it as a hobby now
09:41 PM Tom_test: i've done programming and operating
09:41 PM memleak: ladder logic programming?
09:41 PM Tom_test: gcode
09:41 PM memleak: oh that
09:42 PM memleak: gcode is like assembly to me
09:42 PM Tom_test: ok i renamed the config file directory so when linuxcnc started it asked to save the sim config
09:42 PM Tom_test: it loads the .ngc file fine still
09:43 PM memleak: you sure it's not 3d CHIPS?
09:43 PM Tom_test: i helped him wire up his machines and did some programming for him initially
09:44 PM Tom_test: it shows me 3 circles showing the axis names in each
09:44 PM Tom_test: no it's not chips
09:44 PM memleak: ah ok :)
09:44 PM memleak: well lucky me hahah
09:44 PM Tom_test: heh
09:44 PM Tom_test: gcode is really pretty easy
09:45 PM Tom_test: some parts make it look complex
09:45 PM memleak: i've been told! i could get it down if i worked with it, but it'd probably be of more use to me if i knew anything about mechanical engineering haha
09:46 PM Tom_test: i'm no engineer
09:46 PM Tom_test: just like to learn
09:46 PM memleak: i can rebuild engines aside from honing but that's the most of my abilities
09:47 PM Tom_test: i've done a bit of that in a previous life too
09:47 PM memleak: i thought i was the only one here who wasn't an engineer haha
09:49 PM Tom_test: my friend was a lead at an aircraft plant. quit. opened a shop and i purchased a cad cam package and learned it and started programming for him after wiring his cnc's up to a pc
09:49 PM Tom_test: learning on the fly
09:50 PM memleak: just how bad are the open source cad applications? i've heard they're pretty lousy compared to solidworks but are they of any use to anything?
09:50 PM Tom_test: not good
09:51 PM Tom_test: i'm glad i've got what i have
09:51 PM Tom_test: they get 2d stuff done ok
09:51 PM Tom_test: all have a learning curve
09:51 PM Tom_test: fusion 360 is used by some because of the free licensing of it
09:51 PM Tom_test: acad product
09:52 PM Tom_test: i think they're just trying to gain market share
09:54 PM Tom_test: solidworks & catia are both owned by dassault systems
09:54 PM memleak: autodesk already runs the cad business in terms of gaming, especially because of 3ds max
09:54 PM Tom_test: i've taken some catia classes and like it but it's rather expensive
09:55 PM Tom_test: my kids have taken all of them i believe
09:55 PM Tom_test: catia, autocad, solidworks, some mastercam
09:56 PM Tom_test: for smaller shops, mastercam is probably a good choice
09:56 PM Tom_test: it was a contender when i bought smartcam
09:56 PM Tom_test: (3 decades ago)
09:57 PM memleak: wow!
09:57 PM Tom_test: i have a DOS version still :)
09:57 PM memleak: must have taken forever to render things haha
09:57 PM memleak: come back 3 weeks later
09:57 PM Tom_test: it's pretty quick actually
09:57 PM Tom_test: it's not solids but wireframe
09:58 PM Tom_test: i can load a whole wing section and doesn't take that long
09:58 PM Tom_test: all the new software is solids based
09:59 PM memleak: right, i was thinking it was solids back then but it was all just wireframe, i didn't know that
09:59 PM Tom_test: i can still do full 3d parts
10:00 PM memleak: i don't think any of the new cad software supports radeon gpus though, all cuda..
10:00 PM Tom_test: couldn't say about that
10:01 PM Tom_test: it's getting slower now that they're all going cloud based
10:01 PM memleak: yeah adobe software is the same way
10:02 PM memleak: i hate it.. i hate it all
10:02 PM Tom_test: i'm not fond of adobe
10:02 PM memleak: well neither am i, but your choices really are sony vegas or premiere
10:03 PM memleak: i hated adobe since the 90s with flash constantly crashing and sucking up 100% of memory
10:04 PM Tom_test: shuting this one down. MB is sitting on a cardboard box with the ssd hanging off the edge
10:04 PM memleak: heh!
10:05 PM Tom_L: andy should be happy tomorrow
10:10 PM memleak: yes indeed.
10:10 PM memleak: i'll add the bios setting notes to a README.BIOS file
10:10 PM Tom_L: this was a gigabyte MB fwiw
10:10 PM memleak: or maybe BIOS.README so people don't think it's a bios file..
10:11 PM memleak: gigabyte and asus are the best brands for linux i've seen
10:11 PM memleak: MSI is hit or miss, biostar should be avoided at all costs.
10:12 PM memleak: asrock is pretty bad too.
10:12 PM Tom_L: i've got a couple of those
10:12 PM Tom_L: i may be running wheezy on one
10:13 PM Tom_L: i'd have to raise the hood to be sure
10:19 PM memleak: I'm glad you guys are moving away from wheezy because that's a major security risk right now
10:20 PM Tom_L: heh i use it on my mill still :)
10:20 PM memleak: connected to the internet?
10:20 PM Tom_L: i wonder how well stretch RTAI would do on that hardware
10:20 PM Tom_L: yes
10:21 PM Tom_L: it's not on that much
10:22 PM memleak: stretch RTAI meaning andy's debs?
10:22 PM Tom_L: yeah
10:23 PM Tom_L: probably not that well
10:23 PM Tom_L: i could take the ssd out and see
10:24 PM Tom_L: another day
10:25 PM memleak: i have no idea, the new rtai code i've been working on so far works on one of my AMD APU's, not that great on Ryzen, and both of your guys' systems
10:25 PM memleak: *APUs
10:39 PM Tom_test: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/RTAI_Latency_test_suite_J1900.txt
10:40 PM Tom_test: how do i interrogate the cpu or system data?
10:40 PM Tom_test: i'm not sure what board this is
10:40 PM memleak: oh dmidecode probably
10:40 PM memleak: and lspci for everything else
10:41 PM memleak: lscpu or cat /proc/cpuinfo for CPU stuff
10:42 PM Tom_test: model name : Intel(R) Atom(TM) CPU D525 @ 1.80GHz
10:43 PM memleak: sudo dmidecode -q | head -n 50
10:44 PM Tom_test: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/latency_test.png
10:44 PM memleak: Base Board Information
10:44 PM Tom_test: wrong one
10:44 PM Tom_test: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/latency_D525_stretchRTAI.png
10:45 PM Tom_test: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/RTAI_Latency_test_suite_D525.txt
10:45 PM Tom_test: not horrible
10:45 PM memleak: so 4 good systems, and 1 mediocre one
10:46 PM Tom_test: this is the old intel D525 board
10:50 PM Tom_L: it would probably work ok on the D525, i'm running mesa hardware on it but does it support parallel port?
11:01 PM memleak: does what? the kernel?
11:01 PM memleak: the debs are based off of vanilla debian kernel config files
11:04 PM Tom_L: ok one more board to add to the tests...
11:05 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/Q1900M_Latency_test.png
11:05 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/RTAI_TEST/RTAO_latency_test_suite_AS_ROCK_Q1900M.txt
11:05 PM memleak: good good!
11:05 PM memleak: so that's 3 boards on your end that it works with?
11:06 PM Tom_L: so that's the 3 i have
11:06 PM memleak: all different motherboards and cpus?
11:06 PM Tom_L: yes
11:06 PM Tom_L: first was the gigabyte, 2nd the D525, and the last the ASROCK Q1900M
11:07 PM Tom_L: i've got a couple more but they'd be pretty close to these i think
11:07 PM Tom_L: one is another gigabyte board and the other is an itx version of the Q1900
11:09 PM memleak: that should be good enough!
11:09 PM Tom_L: fair variety of boards
11:09 PM memleak: yes indeed!
11:10 PM memleak: thank you so much for testing these, goes to show that the problem is ryzen, not RTAI
11:10 PM Tom_L: glad to help
11:10 PM memleak: i was even having overruns on this systems, which is very bad
11:10 PM memleak: i'm chasing ghosts
11:11 PM Tom_L: those are AMD chips?
11:11 PM memleak: Ryzen is AMD's latest CPU family, from the Zen architecture
11:11 PM memleak: Post FX-series CPUs
11:11 PM Tom_L: i don't try to keep up anymore :D
11:12 PM memleak: hah, yeah the latest stuff from AMD is all Ryzen, even the Fusion APUs are all Ryzen-based now
11:12 PM memleak: They have a 16-core coming out in September or so, with 70MB L2+L3 cache
11:13 PM memleak: or even 77MB I forget, but it's a lot
11:13 PM Tom_L: on the last 2 tests i have no idea what the BIOS settings were so they could possibly be improved
11:14 PM Tom_L: on the D525 & Q1900
11:14 PM memleak: if they ran preempt_rt or rtai previously, then the bios is probably fine
11:14 PM Tom_L: neither have, they're just test machines
11:15 PM Tom_L: well, not on cnc hardware anyway
11:21 PM memleak: ah, well thank you very much for your help, i'd never know how stable these are if it weren't for you guys
11:23 PM Tom_L: np
11:27 PM memleak: there's yet another ipipe kernel update for 4.14 with core functional changes -_-
11:27 PM memleak: i'm going to hold off on any kernel updates until the Xenomai guys stop messing with the 4.14 kernel
11:28 PM Tom_L: lemme know if i can help test again
11:28 PM memleak: well the only changes i was going to make long-term was add 4.19 support but if they're pushing patches every few days, i might as well just wait
11:28 PM memleak: i'll probably ask for your help in a few weeks :)
11:29 PM Tom_L: true
11:30 PM memleak: after they patch it a few more times, it's like every month or two there's an update
11:30 PM memleak: going from 4.9 to 4.14 introduced a lot of changes which really messed with their heads
11:31 PM memleak: then 4.19 came around and they knew exactly what was going on.