Query: is there an official location for jiffies per second value?

Previous thread: [PATCH] x86: x86_{phys,virt}_bits field also for i386 (v3) by Jan Beulich on Thursday, September 18, 2008 - 12:13 am. (17 messages)

Next thread: Re: How how latent should non-preemptive scheduling be? by Sitsofe Wheeler on Thursday, September 18, 2008 - 12:26 am. (1 message)
From: Grant Coady
Date: Thursday, September 18, 2008 - 12:23 am

Hi there,

I found I can get jiffies per second with this on one box:

grant@deltree:~$ uname -r
2.6.26.5a

grant@deltree:~$ awk 'FNR==NR{up=$1};/^jiffies:/{print up, $2, $2/up}' \
	/proc/uptime /proc/timer_list
420389.04 126026727 299.786


But this value is very different on another box:

grant@pooh64:~$ uname -r
2.6.26.5a

grant@pooh64:~$ awk 'FNR==NR{up=$1}/^jiffies:/{print up, $2, $2/up}' \
	/proc/uptime /proc/timer_list
73830.36 4317026409 58472.2
73830.36 4317026409 58472.2

Is there a reliable way to determine this jiffies value across different 
x86 boxes?

The 'sysconf(_SC_CLK_TCK)' doesn't return the right answer either.  It 
says 100Hz when it is 300Hz.

Thanks,
Grant.
--

Previous thread: [PATCH] x86: x86_{phys,virt}_bits field also for i386 (v3) by Jan Beulich on Thursday, September 18, 2008 - 12:13 am. (17 messages)

Next thread: Re: How how latent should non-preemptive scheduling be? by Sitsofe Wheeler on Thursday, September 18, 2008 - 12:26 am. (1 message)