On Sun, 2004-08-08 at 00:04, Florin Andrei wrote: > On Fri, 2004-08-06 at 22:24, Lee Revell wrote: > > > As Ingo's patches stabilize, it is very important for more people to > > test them. With 2.6.8+voluntary-preempt-O3, using voluntary_preempt=3, > > kernel_preemption=1, and the soundcard and RTC irqs non-threaded, it > > works for me (tm) - I have been unable to produce a latency over 50 > > usecs, even while punishing the machine with parallel builds and iozone > > -a. > > What is the reason to run the RTC IRQ non-threaded? > > In general, why run certain IRQs non-threaded? This ensures that those interrupts have the highest priority of anything on the system. The soundcard is important for obvious reasons, and the RTC is used for midi timing. These are also the only two devices on the system (other than the timer irq 0 which cannot be made threaded) whose nature imposes an inherehent real time constraint on the system. For any other device that can raise an interrupt, we can process it at our leisure, but if we don't handle the sound card interrupt within the period time we are hosed. In practice I have found that if I do not make the rtc interrupt non-threaded I get 'rtc: lost some interrupts at 1024 hz' messages when using mplayer, and xruns in alsa unless i do the same for the sound card. Lee
