Date: Sat, 29 Oct 2005 00:46:24 +1300 From: Mark Kirkwood <markir@paradise.net.nz> To: Poul-Henning Kamp <phk@phk.freebsd.dk> Cc: David Xu <davidxu@freebsd.org>, "Yuriy N. Shkandybin" <jura@networks.ru>, current@freebsd.org Subject: Re: Timers and timing, was: MySQL Performance 6.0rc1 Message-ID: <43620F90.1020401@paradise.net.nz> In-Reply-To: <28339.1130477343@critter.freebsd.dk> References: <28339.1130477343@critter.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
Poul-Henning Kamp wrote:
> In message <43615BBB.2080702@paradise.net.nz>, Mark Kirkwood writes:
>
>>Chuck Swiger wrote:
>>
>>
>>>FreeBSD 5.4-STABLE i386
>>> null function: 0.01278
>>> getpid(): 0.51329
>>> time(): 2.54771
>>> gettimeofday(): 2.54982
>>>
>>
>>Chuck - very interesting results. I happen to have a PIII 1Ghz running
>>5.4-RELEASE so thought it would be interesting to reproduce your
>>numbers. My null and getpid pretty much do, but the time functions seem
>>much quicker on my machine - some sort of regression in 5.4-STABLE maybe?
>
>
> No, different timecounter hardware.
>
> Use
> sysctl kern.timecounter
> to see what your hardware uses.
>
Excellent - thanks for clearing that up that confusion (err... that I
was helping to propagate ...sorry).
e.g: on the 2xPIII 1Ghz Tyan (6.0-RC1):
$ sysctl -w kern.timecounter.hardware=TSC
$ ./timer
null function: 0.01140
getpid(): 0.51597
time(): 0.80246
gettimeofday(): 0.74953
$ sysctl -w kern.timecounter.hardware=i8254
$ ./timer
null function: 0.01152
getpid(): 0.52081
time(): 5.26879
gettimeofday(): 5.23759
Cheers
Mark
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?43620F90.1020401>