Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 25 July 2012 03:54, John Stewart alex.stew...@crc.ca wrote: Yes - e.g., top shows 1 processor. top always shows one processor unless you press the 1 key. Even with Isolcpus top should show 2 processors, but one shold appear completely idle. -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Andy; top always shows one processor unless you press the 1 key. Even with Isolcpus top should show 2 processors, but one shold appear completely idle. Thank you. I'll check this tonight. John Stewart. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Me again; Ok - top (with key press 1) shows 2 cpus, one idle ALL the time. With the LinuxCNC 2.5 release-version, downloaded about a month ago, The taskset is not getting called as it should be. So, LinuxCNC runs on CPU0, along with everything else. Now to find out where that taskset should be getting called from. At least I can turn on ethernet on my linuxcnc computer, because the issue is getting close to being resolved. This stuff should be put in the Troubleshooting document - following Kent's instructions (as great as they are) does not get one to the solution. Thanks; John Stewart. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 25 July 2012 22:47, John Stewart alex.stew...@crc.ca wrote: Me again; Ok - top (with key press 1) shows 2 cpus, one idle ALL the time. I _think_ that means that RTAPI is running on there, but top can't see it. -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Andy; Ok - top (with key press 1) shows 2 cpus, one idle ALL the time. Nope - (being blunt) Remember that that isolcpus is theoretically set properly by grub2. 1) start linuxcnc 2.5.0 from command line. f50mm g1 x0 y0 z0 top shows: cpu0 64% system 35% idle cpu1 100% idle. 2) taskset 0x0002 linuxcnc f50mm g1 x0 y0 z0 top shows: cpu0 70% user 27% idle. cpu1 65% system 33% idle -- I'm going to run some tests tonight with linuxcnc tasksetted to cpu1. Wonder, though, why cpu0 70% user space processing is happening with taskset set? Maybe a tight loop in there waiting for something to happen. I'll report back tomorrow on how well my machine runs tests using an explicit taskset. John Stewart. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 25 July 2012 23:15, John Stewart alex.stew...@crc.ca wrote: 2) taskset 0x0002 linuxcnc Shouldn't that be rtai? Doesn't linuxcnc include all the userspace stuff too? I have never seen anything rtai show up in top (or htop, having just installed that) and I have been assuming it was running invisibly in the apparently empty CPU core. I might, of course, be entirely wrong in this assumption -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Andy; 2) taskset 0x0002 linuxcnc Shouldn't that be rtai? Ah - I now understand what your thinking is. I really don't know - but your comment makes sense - I was throwing the whole of linuxcnc over to cpu1, not just the rtai section. Right now my little KX1 is running a test with a BASE_THREAD of 25000; before it was 30379 and it would throw out the Real Time error within a few minutes. Only knowing about the taskset command for the last few hours, and not knowing how linuxcnc does things, I'm learning as we go along. We'll see how this BASE_THREAD of 25000 test goes - even if the whole of LinuxCNC is on one cpu, as long as it seems to work without croaking, maybe that's all that is needed for my config. JohnS. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On Wednesday 25 July 2012 21:09:02 John Stewart did opine: Me again; Ok - top (with key press 1) shows 2 cpus, one idle ALL the time. With the LinuxCNC 2.5 release-version, downloaded about a month ago, The taskset is not getting called as it should be. So, LinuxCNC runs on CPU0, along with everything else. Now to find out where that taskset should be getting called from. At least I can turn on ethernet on my linuxcnc computer, because the issue is getting close to being resolved. This stuff should be put in the Troubleshooting document - following Kent's instructions (as great as they are) does not get one to the solution. Thanks; John Stewart. Sounds like you have it John. As a side note, rtai looks for that idle cpu and uses it for all the realtime stuff. But because of the isolcpus, it is so isolated that top, htop, gkrellm etc, cannot access it to see how busy it is so it always _looks_ idle. But it isn't idle when linuxcnc is running. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users Cheers, Gene -- There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) My web page: http://coyoteden.dyndns-free.com:85/gene is up! It is the quality rather than the quantity that matters. -- Lucius Annaeus Seneca (4 B.C. - A.D. 65) -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 7/25/2012 5:47 PM, John Stewart wrote: Me again; Ok - top (with key press 1) shows 2 cpus, one idle ALL the time. With the LinuxCNC 2.5 release-version, downloaded about a month ago, The taskset is not getting called as it should be. So, LinuxCNC runs on CPU0, along with everything else. Now to find out where that taskset should be getting called from. At least I can turn on ethernet on my linuxcnc computer, because the issue is getting close to being resolved. This stuff should be put in the Troubleshooting document - following Kent's instructions (as great as they are) does not get one to the solution. Thanks; John Stewart. John, I'm not sure yet what all this stuff is, but certainly I intended my contribution to the wiki to be amended as needed. The fact that the realtime kernel runs beyond (or below, if you like) the ken of the Linux kernel and all its useful tools like top and htop must be taken into account. Good luck. Regards, Kent -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Kent; It looks like I was wrong, but (if I'm correct this time!) running linuxcnc + the RTAI stuff in core1 on my little D525MW seems to do the job quite well. It's been running a program for a couple of hours right now, and no issues. Q: Why, if there was nothing other than RTAI running on cpu1, did I get real time errors, but adding linuxcnc to cpu1 made it run better? Where, really, is the timing issue? Thank all of you for the replies and support and guidance. John Stewart. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 7/25/2012 10:50 PM, John Stewart wrote: Kent; It looks like I was wrong, but (if I'm correct this time!) running linuxcnc + the RTAI stuff in core1 on my little D525MW seems to do the job quite well. It's been running a program for a couple of hours right now, and no issues. Q: Why, if there was nothing other than RTAI running on cpu1, did I get real time errors, but adding linuxcnc to cpu1 made it run better? Where, really, is the timing issue? Thank all of you for the replies and support and guidance. John Stewart. I suspect the path to a satisfying answer lies in understanding the bottom of the page in http://wiki.linuxcnc.org/cgi-bin/wiki.pl?RealTime which mentions the improvement in latency results obtained when running a cpu hog to prevent flushing RT code out of the cache. This is beyond the scope of my wiki entry, which addressed the narrow point of moving from classic Grub to Grub2. Honestly, though, I haven't thought about multi-core issues since I wrote my wiki entry 14 months ago, and I've already forgotten half of what I learned in the process. (Which makes me wonder why I'm trying to understand the issue of USB badness now. How long will it take me to forget whatever I might learn?) Regards, Kent -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On 24 July 2012 03:09, John Stewart alex.stew...@crc.ca wrote: I've got a D525mw, and every once in a while it'll throw up an RTAI error. I found that my D510 was very reliable for latency unless I had the webcam plugged in. -- atp If you can't fix it, you don't own it. http://www.ifixit.com/Manifesto -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Gene, Andy: Andy; I thought it was reliable, too…. (read on) I have 2 of them, running at 23000 ns for a BASE_PERIOD. I can run most of the day without encountering a realtime error. Good - I'll try your BASE_PERIOD - (read on) I assume you also have the option set in the kernel line of your grub.conf Thanks for the slap on the back of the head . Let me check that - I did but it might have been removed by some machine updates. Going to 23000 will make a noticeable difference in the machines top speed in that event. :-) That also assumes you have stepper voltages high enough to make decent torque at those speeds. 24 is often borderline IMO, my mill is running a 28 volt supply, my lathe a 38 volt under load. Running at circa 42V - G540, calculated optimal voltages from equations on gecko web site, and tuned my power supply to that. It's a little Sieg KX1 with a Gecko G540. I'll verify that isoclplus (name like that) parameter, and try your BASE_PERIOD and email back in a couple of days with results. Thanks; JohnS. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On Tuesday 24 July 2012 13:11:56 John Stewart did opine: Gene, Andy: Andy; I thought it was reliable, too…. (read on) I have 2 of them, running at 23000 ns for a BASE_PERIOD. I can run most of the day without encountering a realtime error. Good - I'll try your BASE_PERIOD - (read on) I assume you also have the option set in the kernel line of your grub.conf Thanks for the slap on the back of the head . Let me check that - I did but it might have been removed by some machine updates. And don't forget to turn off that time waster in the bios called hyperthreading. A quick test for that is to check and see how many cpu's the machine _thinks_ it has, running htop for instance. If its not been disabled, it will show 4 cores, but the machine is actually faster with that turned off as the bios code wastes 10% of the machines cycles doing needless context swaps. A solution of sorts that when running linux, will never find the problem it is supposed to fix. Going to 23000 will make a noticeable difference in the machines top speed in that event. : :-) : That also assumes you have stepper voltages high enough to make decent torque at those speeds. 24 is often borderline IMO, my mill is running a 28 volt supply, my lathe a 38 volt under load. Running at circa 42V - G540, calculated optimal voltages from equations on gecko web site, and tuned my power supply to that. Custom wound transformers? It's a little Sieg KX1 with a Gecko G540. I'll verify that isoclplus (name like that) parameter, and try your BASE_PERIOD and email back in a couple of days with results. Thanks; JohnS. NP :) Cheers, Gene -- There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) My web page: http://coyoteden.dyndns-free.com:85/gene is up! I tripped over a hole that was sticking up out of the ground. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
I have just bought a few of these, out the box (LCNC 2.4.3 and 2.5) I am getting jitter in the 5000ns range, are you saying that without the hyper threading turned off, the jitter is really much higher? I haven't connected a motor to see test it and I don't get any RT errors while playing. Greg - Original Message - From: Gene Heskett ghesk...@wdtv.com To: emc-users@lists.sourceforge.net Sent: Tuesday, July 24, 2012 7:18 PM Subject: Re: [Emc-users] BASE_PERIOD and Intel D525MW boards On Tuesday 24 July 2012 13:11:56 John Stewart did opine: Gene, Andy: Andy; I thought it was reliable, too…. (read on) I have 2 of them, running at 23000 ns for a BASE_PERIOD. I can run most of the day without encountering a realtime error. Good - I'll try your BASE_PERIOD - (read on) I assume you also have the option set in the kernel line of your grub.conf Thanks for the slap on the back of the head . Let me check that - I did but it might have been removed by some machine updates. And don't forget to turn off that time waster in the bios called hyperthreading. A quick test for that is to check and see how many cpu's the machine _thinks_ it has, running htop for instance. If its not been disabled, it will show 4 cores, but the machine is actually faster with that turned off as the bios code wastes 10% of the machines cycles doing needless context swaps. A solution of sorts that when running linux, will never find the problem it is supposed to fix. Going to 23000 will make a noticeable difference in the machines top speed in that event. : :-) : That also assumes you have stepper voltages high enough to make decent torque at those speeds. 24 is often borderline IMO, my mill is running a 28 volt supply, my lathe a 38 volt under load. Running at circa 42V - G540, calculated optimal voltages from equations on gecko web site, and tuned my power supply to that. Custom wound transformers? It's a little Sieg KX1 with a Gecko G540. I'll verify that isoclplus (name like that) parameter, and try your BASE_PERIOD and email back in a couple of days with results. Thanks; JohnS. NP :) Cheers, Gene -- There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) My web page: http://coyoteden.dyndns-free.com:85/gene is up! I tripped over a hole that was sticking up out of the ground. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users No virus found in this incoming message. Checked by AVG - www.avg.com Version: 8.5.455 / Virus Database: 271.1.1/5144 - Release Date: 07/20/12 18:17:00 -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On Tuesday 24 July 2012 15:28:01 Belli Button did opine: I have just bought a few of these, out the box (LCNC 2.4.3 and 2.5) I am getting jitter in the 5000ns range, are you saying that without the hyper threading turned off, the jitter is really much higher? I haven't connected a motor to see test it and I don't get any RT errors while playing. Greg Even 5000 seems a tad high. Freshly booted, I get about 2300 and it takes some serious work on another screen to run it above 3000. These are the completed boxes labeled OAK, shoebox style with a dvd writer included. $249+ship each. Cheers, Gene -- There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) My web page: http://coyoteden.dyndns-free.com:85/gene is up! Win NT error 001: Error recording error codes. All further errors not displayed. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
Gene (and, others) I assume you also have the option set in the kernel line of your grub.conf (or menu.lst, depending on the mood of the distro divided by the phase of the moon) isolcpus=1, Yes - e.g., top shows 1 processor. When running a Gcode program, it's averaging 68% idle. Hyperthreading is off in BIOS, as is audio, ethernet, and serial port. I do have in the machine: 2nd PCI serial port; for cnc4pc mpg; touch probe code in the HAL file (cnc4pc touch probe) I did bring the BASE_PERIOD down from 30,000 to 28,000; last time a real time error came up was when moving the USB mouse over the linuxcnc window border. Poof! up comes the real time error message. I guess I'll just live with it; it does take about 10 minutes for it to show up, so it's not too bad (I guess??) Thanks; John Stewart. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] BASE_PERIOD and Intel D525MW boards
On Monday 23 July 2012 22:16:53 John Stewart did opine: I've got a D525mw, and every once in a while it'll throw up an RTAI error. At first I thought it was the update manager, but turning that off did not help. my BASE_PERIOD is 30769; I wonder if that is way off the mark? Anyone with an Intel D525MW willing to tell me what you use? I have 2 of them, running at 23000 ns for a BASE_PERIOD. I can run most of the day without encountering a realtime error. I assume you also have the option set in the kernel line of your grub.conf (or menu.lst, depending on the mood of the distro divided by the phase of the moon) isolcpus=1, which reserves the 2nd core for the realtime stuff, and that you have the Hyperthreading option in the bios disabled as that will tear up the validity of the latency reported. (I'll admit to not knowing *which* latency number to choose from the stepconf screens, but, my mill seemed to work…) It should, bearing mind mind that 3 may restrict your choices of microstep rates that are actually usable to the lower values of 8 and under. Going to 23000 will make a noticeable difference in the machines top speed in that event. That also assumes you have stepper voltages high enough to make decent torque at those speeds. 24 is often borderline IMO, my mill is running a 28 volt supply, my lathe a 38 volt under load. Cheers, Gene -- There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) My web page: http://coyoteden.dyndns-free.com:85/gene is up! Sorry, no fortune this time. -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
