On 9/23/2012 3:34 AM, Michael Haberler wrote: > I just stumbled upon an idea (yet untested) to improve latency with > RT_PREEMPT on multicore cpu's like the atom > > this assumes that rtapi_app in sim mode (plus hal_parport in usermode) will > be the time-critical component > it borrows from the isolcpus=1 idea for RTAI where CPU1 would be reserved for > the base thread > the idea is to do something similar for rtapi_app: > > quote from: Max Krasnyansky's post in > http://www.digipedia.pl/usenet/thread/19080/11385/ > >> Actually this is kind of available right now with vanilla 2.6 kernel. >> I'm talking about CPU reservation. Here is an example. >> Let's say we have dual CPU box and we want to dedicate CPU 1 to >> our application: >> - Configure the kernel with following boot options: >> isolcpus=1 acpi_irq_nobalance noirqbalance >> >> This excludes CPU 1 from the scheduler balancing logic. And disables >> ACPI and SW irq balancing. >> Make sure that you don't run user-space IRQ balancer. >> >> - Redirect all interrupts to CPU 0 >> for i in /proc/irq/*; do >> echo 1 > $i/smp_affinity; >> done >> >> - Your app call now migrate to CPU 1 >> int cpu = 1; >> uint32_t mask = (1 << cpu); >> sched_setaffinity(0, sizeof(mask), (cpu_set_t *) &mask); >> >> That's it. CPU 1 is yours. There will be almost zero activity on it, besides >> your task of course. >> >> Max > what would needed to be done is to add the sched_setaffinity() code fragment > to rtapi_app and do the boot/setup incantations before starting linuxcnc > > - Michael > > <deleted prior messages in thread>
Thanks, Michael. Despite my waspish comment about custom-building my own kernel, I found myself doing just that, mostly because I know I'll have to relearn the process anyway to re-purpose some ARM appliances I have in my toybox. Nothing I've done so far in terms of turning off or deleting features from the kernel has made any noticeable improvement over the distributed Debian RT kernel, but of course I'm working in profound ignorance of what's going on inside the Atom processor and the ASUS BIOS. I feel like Rutherford trying to deduce the nature of atoms by shooting neutrons into a metal foil. (Yes, the pun is intended; I'm a physicist after all!) Regards, Kent ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://ad.doubleclick.net/clk;258768047;13503038;j? http://info.appdynamics.com/FreeJavaPerformanceDownload.html _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
