[LAU] resubmitting questions from last week

Brent Busby brent at keycorner.org
Tue Aug 4 14:35:24 EDT 2009


On Mon, 3 Aug 2009, Paul Neaveill wrote:

> Sorry if someone responded and we missed it, but despite much googling 
> and all of that, am still working on getting my own kernel(s) [2 
> machines] going with the following as had been suggested last week:
>
> preempt_RCU
> No_Hz
> HZ_1000
>
> and the other Rock-solid low-latency audio tweaks
> 99 ff migration
> 99 ff posixcputmr
> 98 ff IRQ-8 (real-time clock)
> 97 ff audio IRQ
> 80 rr Jack
>
> Found the preempt_RCU and HZ_1000 spots in the debian/ubuntu kernel 
> hack, still looking for the rest please.?
>
> ** Would really appreciate the assistance with figuring out where/how 
> to put those.

I never did get an answer to some of those questions either, so I 
figured maybe I'd been asking too many.  (I have been posting a lot of 
them lately.)

I think the ones you've got at the top for kernel options should be 
safe.  I'm going to try -- at least initially -- not adjusting any IRQ 
priorities though.  It might work, considering I've got a fast machine, 
and from what I hear, there are even people getting away with ordinary 
non-RT kernels these days.

Preempt RCU was supposed to be buggy for awhile, but tested now for two 
years...  (Source:  LKML)

Somebody said NO_HZ was actually a good thing on a laptop being used for 
pro-audio, and I suppose if you can do well on a laptop you can do well 
anywhere...  (Source:  Some guy with a laptop.)

And HZ_1000 is recommended almost everywhere.

I don't know what the meaning of NO_HZ and HZ_1000 is together, but I 
guess it'd mean a jiffy rate of 1000Hz, but only when interrupts are 
needed?

I'm just putting stuff together from lots of googling too...

-- 
+ Brent A. Busby	 + "We've all heard that a million monkeys
+ UNIX Systems Admin	 +  banging on a million typewriters will
+ University of Chicago	 +  eventually reproduce the entire works of
+ Physical Sciences Div. +  Shakespeare.  Now, thanks to the Internet,
+ James Franck Institute +  we know this is not true." -Robert Wilensky



More information about the Linux-audio-user mailing list