[ntp:hackers] timespec vs timeval

Poul-Henning Kamp phk at phk.freebsd.dk
Tue Jul 17 16:46:38 UTC 2007


In message <469CF153.3080600 at udel.edu>, "David L. Mills" writes:

>The bottom line is that it 
>takes half a microsecond to read the kernel clock in Solaris 10 with a 
>Blade 1500. It's hard to see the interrupt latency is comparable or 
>longer than that.

I hate tell you, but that is far from the case, at least with network
chips.

Almost all recent network chips wanders around in memory for ages
before they issue an interrupt.

As long as the packets you want to time are the only ones, and never
arrive back to back and the system is not anywhere close to CPU bound,
then you can get good low jitter on network packets.

As soon as one of those assumptions are not fulfilled, all bets are
off.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.


More information about the hackers mailing list