[Info-vax] $SETIMR granularity

[SeanOBanion]

On Oct 24, 7:49 pm, Snowshoe <n... at spam.please> wrote:
> Does $SETIMR have the same (1/100 second) granularity on Itanium and
> Alpha as it does on the VAX?  Or is it 1/1024 second due to the clock speed?
>
> (if 1/100 second, how is this done, 1024 doesn't divide 100 evenly)

The 1024 Hz is the interval clock for Alpha systems: not from the
Alpha chip but the support chip timing which always derived  a 1024
interrupt.  The Alpha Architecture had I think 3 or 4 bits that
allowed other clock rates, but only the first, 1024 Hz, timing was
implemented.

Can't find the reference I recall, but Google seems to find
http://h71000.www7.hp.com/wizard/wiz_8572.html
" Various Alpha systems have a hardware tick rate of 1024 interrupts
per second."
Along with some other interesting issues.

I tripped over this when adding an Alpha to a VAX ELN realtime system,
where the control and analysis code initially assumed a 1000 Hz rate
on the Alpha (compared to 100 Hz for the VAX...), but we quickly saw
that wasn't true, and some research found the 1024 Hz rate.  Once we
know that, it was easy to work with...



Sean