Back
[02:18:10] <KGB-linuxcnc> 03Norbert Schechner 05master 235bc99 06linuxcnc 10share/gscreen/skins/gmoccapy/gmoccapy_handler.py 10share/gscreen/skins/gmoccapy/release_notes.txt gmoccapy_0_9_9_9_3 - feed override bug solved * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=235bc99
[02:18:10] <KGB-linuxcnc> 03Norbert Schechner 05master 19ce87b 06linuxcnc Merge branch 'master' of ssh://norbert@git.linuxcnc.org/git/linuxcnc.git * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=19ce87b
[07:12:06] <skunkworks> 2.4GHZ/747000 clocks - invert - 311us total thread time..
[07:14:12] <skunkworks> that is about half - awesome
[07:35:21] <skunkworks> and look at that - following error is about tice as good.. .00016 vs .00031
[07:35:30] <skunkworks> http://imagebin.org/290113
[07:35:43] <skunkworks> funny that I seemed to have to increase ff2 just a smidge
[07:44:10] <skunkworks> this machine (think it is a dual core petium 4 2.4ghz seems to run 2khz just fine)
[07:50:08] <skunkworks> Now - after micges changes last night. Great!
[09:04:13] <skunkworks> running for a couple hours now at 2khz
[09:12:14] <skunkworks> was the read time that variable before?
http://imagebin.org/290132
[09:13:01] <cradek> my wheezy/rtai machine ran opengl programs all night and is showing 9877
[09:13:10] <skunkworks> wow - nice
[09:13:44] <skunkworks> ship it!
[09:14:04] <cradek> either I always get lucky, or most machines work fine with rtai, or most machines that are like my available junk work fine with rtai
[09:14:10] <cradek> hard to say which it is
[09:14:27] <skunkworks> I have good luck with rtai
[09:14:49] <archivist> I have had a few terrible PCs
[09:15:18] <skunkworks> the matsuura ran quite a while with the servo fix.. now we need to figure out why dad can't turn the spindle on..
[09:15:22] <archivist> I have one that I think is a special case that may be worth investigating
[09:16:01] <skunkworks> when I was there the spindle came right on.. both time dad tried - it didn't :)
[09:16:20] <cradek> skunkworks: maybe it's related to the toolchanger not working - orient sensor problem maybe?
[09:16:41] <cradek> mine will act that way when the orient pin is sticky (it is sometimes)
[09:16:56] <skunkworks> right - we have to dig into it.. I had fiddled with the tool changer, collet unclamp when I left - I may have left it in an odd state
[09:17:30] <skunkworks> it sound really nice.
[09:17:50] <skunkworks> the servos give quite a squeel when you push them\
[09:18:26] <skunkworks> even if it isn
[09:18:54] <skunkworks> even if it isn't 100% operational - it will make it so we don't have to guess what 90% of the stuff is doing in there to convert
[09:19:23] <cradek> yeah it's so nice to understand the old stuff
[09:19:40] <cradek> but if I had to choose, having schematics is better
[09:19:58] <skunkworks> yes - your jr schematics helped - but it is a little different
[09:20:28] <skunkworks> I have some feelers out to see if I can find atleast an operations manual and if I am really lucky a service manual
[10:47:50] <KGB-linuxcnc> 03Norbert Schechner 05master b16a107 06linuxcnc 10share/gscreen/skins/gmoccapy/gmoccapy_handler.py 10share/gscreen/skins/gmoccapy/release_notes.txt gmoccapy_0_9_9_9_4 - added a pin to delete message * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b16a107
[11:07:34] <skunkworks> pcw_home,
http://imagebin.org/290143
[11:07:58] <skunkworks> 2.7ghz 2 nics - pro/100 is the 7i80
[11:08:13] <skunkworks> around 200us read times
[11:08:37] <skunkworks> I had to go back to .00015 for ff1
[11:08:49] <skunkworks> and I can surf with the other
[11:08:51] <skunkworks> nic
[11:09:03] <skunkworks> amd system
[11:11:33] <skunkworks> 2khz
[11:12:10] <skunkworks> looks like I could reduce ff2 a little more.. (and I meant ff2 above)
[11:13:19] <pcw_home> I cant quite run my desktop at 2 KHz (flash videos cause latency spike that push it over the edge)
[11:13:21] <pcw_home> but the D525 seems better. I suspect the worst spikes can be reduced by using the RX_RING,TX_RING
[11:13:22] <pcw_home> interface (that avoids the system calls entirely)
[11:15:16] <pcw_home> this basically give you memory access the the drivers ring buffers (even polling for the rx packet is just polling the packet header in the ring buffer)
[11:16:24] <pcw_home> s.the the/to the/
[11:17:06] <skunkworks> sounds promising
[13:22:43] <seb_kuzminsky> http://www.anandtech.com/show/7724/it-begins-amd-announces-its-first-arm-based-server-soc-64bit8core-opteron-a1100
[13:23:49] <skunkworks> pcw_home, I had to switch the nics.. - I was getting realtime delays when I had the 7i80 hooked into the pro/100 pci cards - Swapped it so it was hooked to the onboard with the internet hooked into pro/100
[13:24:57] <andypugh> I want to add isolcpus to a u-Boot device.
[13:26:18] <andypugh> It appears that the way to do this is _not_ to setenv bootargs as the Internet insists. That isn't working. Someone on a different channel insists that I need to recompile the kernel. This is no big deal, but I don't know how to configure isolcpus at compile time. Any ideas?
[13:27:29] <skunkworks> pcw_home, and it is running 2khz
[13:28:25] <seb_kuzminsky> andypugh: i have very little experience with uboot, but *surely* it can pass arguments to the kernel it's booting?
[13:28:40] <seb_kuzminsky> it must do things like tell the kernel it's boot device
[13:28:50] <skunkworks> pcw_home, and it plays youtube vidoes.. ;)
[13:28:51] <andypugh> This is why I am seeking a second opinion.
[13:29:42] <seb_kuzminsky> i bet the internet is right, and uboot can pass args to the kernel
[13:33:07] <andypugh> seb_kuzminsky: Does this mean anything to you? The result of interrupting the boot process and typing "printenv"
http://pastebin.com/HEUEmhw3
[13:38:24] <andypugh> Hmm, "mmcargs" looks a lot like the result of cat /proc/cmdline
[13:38:58] <andypugh> Though I have no idea at all how to edit that in a serial terminal....
[13:40:35] <zultron> Agh, that's awful, an ARM-based Opteron?
[13:40:51] <zultron> So sad. I used to work for AMD. They were once awesome.
[13:41:50] <archivist> reusing the opteron name is just wrong
[13:43:55] <zultron> It makes me want to cry.
[13:46:12] <skunkworks> on a random note - they just got kitkat builds for my nook color.. can't wait to try it. kitkat running decent on 2010 technology..
[13:46:26] <skunkworks> those xda people are awesome
[14:29:45] <skunkworks> micges, running well.
[14:40:11] <skunkworks> http://www.youtube.com/watch?v=dpxS1t874DY&feature=youtu.be
[14:40:16] <skunkworks> sorry - crappy video
[14:50:28] <seb_kuzminsky> andypugh: are you using a boot.scr bootscript from your root image?
[14:51:14] <micges> skunkworks: nice
[14:51:29] <micges> skunkworks: hdd didn't fell? ;)
[14:52:46] <skunkworks> wow - that is close to the edge...
[14:53:14] <micges> skunkworks: what is read time?
[14:53:57] <skunkworks> averages at about300us or so with some spikes
[14:54:13] <skunkworks> but it is running no problem at 2khz thread
[14:55:24] <skunkworks> http://imagebin.org/290143
[14:56:12] <micges> great
[14:57:23] <skunkworks> Great work!
[14:59:53] <andypugh> seb_kuzminsky: I really have no idea
[15:00:31] <andypugh> seb_kuzminsky: But I edited mmcargs and now have the expected kernel paameters, and they succesfully crash Xenomai :-/
[15:00:45] <seb_kuzminsky> andypugh: i think the boot command on line 6 is what uboot runs to boot your system
[15:01:13] <andypugh> Yea, that was my guess too.
[15:01:18] <seb_kuzminsky> it tries to load boot.scr from the mmc partition, and if that succeeds, it runs it
[15:01:36] <seb_kuzminsky> if it fails to load the script it runs the internal scripts
[15:01:37] <andypugh> Any idea how to test if isolcpus is working?
[15:01:43] <seb_kuzminsky> errr
[15:01:44] <seb_kuzminsky> no
[15:02:04] <skunkworks> top and see if 1 processor is 0%?
[15:02:15] * skunkworks doesn't really know
[15:02:17] <andypugh> All processors are at 0% :-)
[15:05:38] <PCW> They must be _very_ fast
[15:06:01] <andypugh> It's running a serial terminal and nothing else
[15:06:43] <PCW> unix is always running something else
[15:07:29] <andypugh> Much of today I have not been running Unix, just u-boot
[15:07:31] <skunkworks> andypugh,
http://www.youtube.com/watch?v=qgnChnpW58M
[15:09:51] <andypugh> I prefer this instance of Andy Pugh:
http://www.linkedin.com/pub/andy-pugh/55/372/275
[15:10:22] <skunkworks> wow - I would not have thought it a common name..
[17:37:08] <memleak> andypugh, not even udev?
[20:48:37] <skunkworks> logger[mah]:
[20:48:37] <logger[mah]> skunkworks: Log stored at
http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2014-02-01.html