[Asterisk-Users] Re: ztdummy inaccuracy on linux-2.6

[Tony Mountifield]

In article <43CBCD3F.9050006 at gmail.com>, Tamas <jalsot at gmail.com> wrote:
> Hello,
> 
> I have some ugly numbers given by zttest for ztdummy on an AMD64 box
> running linux-2.6.15 compiled for Athlon64.

Don't be misled by the apparent ugliness of the numbers! Are you hearing
any effects in the audio? If not, then stop worrying!

The output of zttest is misleading, since it is just relying on the kernel
time of day clock for measuring. Because of this, the "jiffies" version will
appear to show more consistent results, and the RTC version will be a little
more variable. But that is not because the RTC version of ztdummy is less
accurate. On the contrary, it is more accurate, on average (8000 calls
out of every 8192 per second). It is the reference for the calculations
that is less accurate.

I'm glad that your post mentioned bug number 5971, as I hadn't been aware
of that one until now. I think the patch in that bug is seriously flawed,
and have just posted on it to that effect.

> [... snip results ...]

> 
> HW:
> Tyan Tomcat K8E, Athlon64 3000+, 1GB RAM, 3ware 8006, 2x Maxtor HDD
> 
> SW:
> Ubuntu 5.10, linux-2.6.15, zaptel from 1.2 branch
> 
> Any idea what can be wrong?

Nothing! Please ignore the figures, and use ztdummy with USE_RTC without
the patch from 5971. Then just describe any operational effect you might
be seeing, if any.

> Thanks in advance,
> 	Tamas

Cheers
Tony

-- 
Tony Mountifield
Work: tony at softins.co.uk - http://www.softins.co.uk
Play: tony at mountifield.org - http://tony.mountifield.org