#linuxcnc-devel | Logs for 2016-04-20

[02:48:11] <seb_kuzminsky> :q
[02:48:20] <seb_kuzminsky> oops
[02:48:22] <seb_kuzminsky> time for bed
[08:07:44] <skunkworks> https://groups.yahoo.com/neo/groups/mach1mach2cnc/conversations/messages/150666
[08:08:35] <skunkworks> pcw_home: thanks - going to try it.
[08:08:49] <skunkworks> almost to the point of moving some axis..
[11:50:51] <JT-Shop> zlog
[11:57:06] <JT-Shop> <pcw_home> The h97 pro4 has been flawless at 4 KHz for about a year uptime
[11:57:29] <JT-Shop> pcw_home: are you running rtai or rt preempt on that one?
[12:09:59] <pcw_home> RT-Preempt
[12:10:11] <lair82> JT-Shop, I'm using that same board, and the questions on the user channel are related to this machine.
[12:10:30] <pcw_home> it has similar RTAI and RT-Preempt latencies (7-10 usec)
[12:10:51] <mozmck> pcw_home: you mentioned you might change the default IP address to at some point in the 7i92 - have you done that?
[12:11:20] <pcw_home> the defualt EEPROM address is
[12:11:27] <pcw_home> default
[12:11:45] <mozmck> And it used to be 192.168.1.x?
[12:12:08] <pcw_home> Yes it used to be
[12:12:41] <pcw_home> I will leave the ROM address (jumpers down) at
[12:12:57] <mozmck> On all the board so far we move a jumper to change the IP to
[12:13:30] <pcw_home> yes that will work (jumpers down,up)
[12:13:57] <pcw_home> you can also use bootp :-)
[12:14:16] <mozmck> Ok, so our jumper settings will stay the same.
[12:15:11] <pcw_home> I want to keep the ROM address since its a in a very common internal network range
[12:15:14] <mozmck> bootp? I think moving a jumper is easier :-)
[12:15:50] <JT-Shop> will a celeron processor be ok for that h97 pro4 mb?
[12:15:59] <mozmck> Ok - I was thinking you said you might change that - doesn't matter to me.
[12:16:10] <pcw_home> well bootp can be set to work for a range of mac addresses so can be made to always work with mesa Ethernet FPGA cards
[12:16:50] <pcw_home> (which always are 00 60 1B XX XX XX )
[12:17:03] <mozmck> I see.
[12:18:34] <pcw_home> you can edit the source if you wish to change the ROM ID
[12:18:35] <pcw_home> I guess I could add 2 ROM addresses (theres an unused jumper setting)
[12:19:51] <mozmck> It doesn't matter to me, I was just checking if it was changing on the new boards. Easy enough for us to move a jumper!
[12:23:54] <pcw_home> we have a customer thats changing the ROM IP address,
[12:23:55] <pcw_home> But I honestly don't think is a great idea at least testability wise
[12:23:57] <pcw_home> (and mesaflash has the as a default IP if not specified)
[13:34:29] <PCW> With the latest Preempt-RT kernel (4.4.7-RT16) and the kernel config I swiped from OSADL
[13:34:31] <PCW> The Zotac CI323 has been up a couple days with hm2_eth at 1 KHz running videos etc
[13:36:23] <PCW> so maybe the 4.4.x kernel are good for something ( they just don't like the Core Duo )
[13:50:31] <JT-Shop> is there a bit file for the 7i92 for a parallel port bob like the C1 or C1G from cnc4pc?
[13:51:18] <PCW> Just tried 4.4.7-rt16 on the G3258 and its fine so maybe the 4.4.x kernels
[13:51:20] <PCW> just cause pain and suffering on older hardware
[13:51:42] <PCW> JT-Shop: probably
[13:52:31] <JT-Shop> looks like they are typical outputs on 1,2,3,4,5,6,7,8,9,14,16,17
[13:52:47] <JT-Shop> in inputs on 10,11,12,13,15
[13:53:10] <PCW> 7i92_DMMBOB1x2D.bit
[13:53:12] <PCW> 7i92_G540x2D.bit
[13:53:14] <PCW> 7i92_MX3660x2D.bit
[13:53:15] <PCW> 7i92_R990x2D.bit
[13:53:17] <PCW> are likely suspects
[13:53:47] <JT-Shop> thanks
[13:53:56] <PCW> you could look at the similarly named 5i25 .pin files for pinouts
[13:55:10] <PCW> or do
[13:55:12] <PCW> mesaflash --device 7i92 --write xxxxx --reload
[13:55:14] <PCW> mesaflash --device 7i92 --readhmid >7i92_xxxx.pin
[13:55:16] <PCW> a bunch of times
[13:56:55] <JT-Shop> the 540 has the right pin out
[13:57:10] <lair82> PCW, would it be worth updating my RT kernel then, if you are seeing favorable results on the 4.4 kernels?
[13:57:54] <PCW> I doubt it, the G3258 has been fine with most kernels post 3.18 or so
[13:58:45] <lair82> Also, should I be able to build Preempt RT uspace on Mint 17.3 using the cheat sheet you sent me to install and update Preempt?
[14:00:51] <lair82> Here is what you sent me, https://app.box.com/s/i3wuqj1e68po2zx57bnehbqmqx7jra0r
[14:01:38] <PCW> yeah I think you need to add ssl-dev or something
[14:01:39] <PCW> I'm running Mint 17.3 on this:
[14:01:41] <PCW> http://www.newegg.com/Product/Product.aspx?Item=N82E16856173128&cm_re=zbox-_-56-173-128-_-Product
[14:01:42] <PCW> not tried anything fancy like gmoccapy though
[14:02:07] <lair82> OK, I will leave well enough alone in regards to the G3258, I'm up for trying anything
[14:02:22] <lair82> Anything to get off of Wheezy
[14:03:33] <lair82> Your running a CNC control with that?????
[14:03:55] <lair82> And I could hook my 7i80 to that thing and run my mill?
[14:04:47] <PCW> yeah since you dont need any slots (and that box has dual LANs)
[14:04:58] <lair82> Thats sick.
[14:05:30] <PCW> its still much slower than the G3258 so probably not the best for fancy systems
[14:06:02] <PCW> but they are nice, and can be mounted on the back of a monitor
[14:06:09] <lair82> Regardless, that is pretty wild, that I could control a CNC with that little thing
[14:07:24] <PCW> you can control a CNC with much less
[14:12:39] <MarkusBec> you can control a cnc with a single 8051
[14:13:53] <lair82> And I do or do not need to use the "isolcpus=1" when running wheezy using Preempt RT?
[14:14:44] <jepler> if you are happy with the latency you get, then don't change that setting.
[14:15:02] <jepler> don't change things you don't need to change
[14:17:02] <lair82> I won't if I don't need to, but I was chasing my tail with RT faults, and the ethtool command was what actually fixed it, so I'm not sure about the isolcpus=1.
[14:18:17] <jepler> on two core or one core, two-thread systems, some people report that isolcpus=1 has a beneficial effect.
[14:18:25] <jepler> the flag does have an effect on rt-preempt systems
[14:18:37] <jepler> but since only you can test your system, I can't tell you whether you benefit from it
[14:22:14] <lair82> How could I revert the Isolcpus=1 to remove it, so the systems functions as if it wasn't there, that way I could test if it is helping or not?
[14:30:56] <PCW> I my experience with the G3258, isolcpus made Ethernet slower
[14:31:35] <PCW> that is, it limited the maximum thread rate I could acheive
[14:35:18] <lair82> How do remove it then?
[14:39:48] <mozmck> Isn't it something you added to the grub boot config? You would remove it from there.
[14:40:17] <mozmck> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?The_Isolcpus_Boot_Parameter_And_GRUB2
[14:40:41] <PCW> yeah, edit /etc/default/grub
[14:42:16] <PCW> remove the offending cmd line tokens and run update-grub
[14:44:28] <mozmck> yeah, forgetting to run update-grub has bitten me a few times.
[14:44:37] <lair82> Aaahhh, I see, ok, thanks
[14:46:26] <lair82> I'm out, talk to you guys later, Thanks
[17:15:35] <JT-Uspace> I just copied and pasted a 4th axis to a config and get this error emcTrajSetAxes failing: axes=4 axismask=47
[17:15:48] <JT-Uspace> emcMotionInit: ini_hal_init fail
[17:16:22] <JT-Uspace> HAL: ERROR: duplicate component name 'inihal'
[17:16:40] <JT-Uspace> ini_hal_init: ERROR: hal_init() failed
[17:20:21] <JT-Uspace> oh the number of axes vs the axis names I had U and 4
[18:26:54] <skunkworks> servos hooked up - estop/overtravel loop figured out. So close to motion.