Re: Clarifications about PPS SHM content

[Paul Theodoropoulos]

On 3/25/2020 19:33 PM, Gary E. Miller wrote:
> Yo Martin!
>
> On Wed, 25 Mar 2020 21:47:31 -0400
> Martin Boissonneault <address@hidden> wrote:
>
> > I've set /ntpheat -c 3 -t 68/ for testing. Does it need to be niced?
> Nope, just the opposite.  The point of ntpheat is waste lots of
> cpu cycles and make heat, when needed.  It will pause, and not
> waste cpu cycles when something else sends the LF up and makes heat.
In my experience, at least on RPI 2 at least, ntpheat introduced too much 
secondary 'noise' for lack of a better term - since rpi2 can only bounce 
between 600Mhz and 1.4Ghz, rather than other hardware that has 
finer-grained governor options. I ended up just putting everything in an 
enclosure with lots of insulation - then fine tuning it to allow enough 
heat to escape to prevent overheat. Still fluctuates over a day of course, 
but much narrower fluctuations.

If you don't have the option of a carefully insulated environment, ntpheat 
is definitely an improvement over free-wheeling temperature changes.

--
Paul Theodoropoulos
www.anastrophe.com