On Thursday 24 October 2013 16:45:45 Kent A. Reed did opine: > On 10/24/2013 11:13 AM, Gene Heskett wrote: > > On Thursday 24 October 2013 11:05:25 Kent Reed did opine: > >> On Thu, Oct 24, 2013 at 10:10 AM, Gene Heskett <ghesk...@wdtv.com> wrote: > >>> <...> > >>> > >>> > >>> Note the isolcpus=1 on the end of the 'linux' line, it has to be > >>> hand edited into that file after the installation is done, and then > >>> rebooted again to enable it. The UUID's should be different from > >>> the above. > >> > >> It's generally a bad idea to hand-edit the grub.cfg file. It even > >> says why at the top just after the "DO NOT EDIT THIS FILE" comment. > >> > >> I posted a page to the Wiki two years ago ( > >> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?The_Isolcpus_Boot_Parameter_ > >> And _GRUB2) about my solution to the problem. > > > > I think I followed this recipe when I did that. It does give some > > extra whitespace in front of the argument. > > > > In this case , since we don't normally update the kernel, the grub.cfg > > rebuild isn't being done, so either method would get the job done. > > Well, there are other reasons. For instance, one decides to save some > space by deleting old kernels or one decides to try out an alternative > kernel, say of the Xenomai or rt_preempt persuasion, on a working > machine. In either case, update-grub is run to make grub aware of the > changes. > > > What I wanted to offer was a quick & dirty way for the OP to check > > that it was done, just by looking at the generated grub.cfg. > > I hear you , Gene, but the phrase "...it has to be hand edited into that > file after the installation is done..." is now enshrined in the email > archives and I didn't want it looking to future readers like a > recommended practice.
Yes, that was a mistake no doubt. I didn't word that well at all. > > My excessive sense of caution stems from 30 years living with myriad > systems and applications, many of which bit me in the backside one time > or another. A quick-n-dirty hack which gets a program going today can > have a nasty habit of remaining in the code to become a landmine for the > unsuspecting tomorrow. > > I don't claim to be guiltless just because I was often a victim. I've > left a few landmines of my own. We all have. :( > > More to the point, I wish people who feel sufficiently compelled to post > numbers to the latency-test wiki page would also feel compelled to run > the tests with these different kernel configurations and post all the > results. See for example, my three entries for the ASUS AT5NM10-O. This > would help reduce the amount of handwaving we do on this list. Well, with those kernel options enabled, my latency figures are indeed excellent. on a pair of D525MW's. There is a realtime warning about once a day, but a latency.test run on either of my boxes gets maybe 3 u-secs of wobble. This is 15 u-s or more better than any other board I've had running my stuff. I was able to cut my base_thread timing in half. It will run at 20 u-s, but feels much better at 23. And the servo thread on the lathe is running at 2khz nominally because it cut the noise in the encoder output down to plumb reasonable amounts. > Just my 2 cents worth. > > Regards, > Kent > > > ------------------------------------------------------------------------ > ------ October Webinars: Code for Performance > Free Intel webinars can help you accelerate application performance. > Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most > from the latest Intel processors and coprocessors. See abstracts and > register > > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.cl > ktrk _______________________________________________ > 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) There is no satisfaction in hanging a man who does not object to it. -- G. B. Shaw A pen in the hand of this president is far more dangerous than 200 million guns in the hands of law-abiding citizens. ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
