On Thu, Aug 11, 2011 at 15:06, Nikos Chantziaras <[email protected]> wrote:
> On 08/11/2011 10:35 AM, Pandu Poluan wrote:
>>
>> On Thu, Aug 11, 2011 at 14:27, Nikos Chantziaras<[email protected]>  wrote:
>>>
>>> On 08/11/2011 09:56 AM, Pandu Poluan wrote:
>>>>
>>>> Just wondering, is it possible to specify a custom timer frequency?
>>>> E.g., HZ=500 instead of one of the canned values (100, 250, 300,
>>>> 1000).
>>>
>>> It is possible, but it's a bad idea because non-standard values can
>>> result
>>> in driver breakage.  Some code assumes specific timer granularities
>>> (100Hz =
>>> 10ms, 250Hz = 4ms, etc).  This usually happens with values above 1000Hz,
>>> so
>>> it might be possible to experiment with non-standard sub-1000Hz values.
>>>
>>> But why do you want a custom value anyway?
>>>
>>
>> Well, for a firewall, I've calculated (gathered and extrapolated from
>> a lot of sources), the latency per-packet is usually less than 1 ms,
>> at worst still less than 2 ms.
>>
>> Thus, setting the timer freq to 100 Hz (as suggested for 'normal'
>> server load) means the timeslice is way too long (10 ms per time
>> slice).
>>
>> So, I speculate that better -- and uniform -- performance will be
>> achieved with a timer freq of 500 Hz.
>>
>> Of course, this is a wild speculation/guess from me. I never quite
>> understand netfilter/xtables' relation to the timeslices, so I may be
>> talking nonsense :-)
>
> This assumes that networking is dependent on the timer interrupt, which
> doesn't seem to be the case; going from 100Hz to 1000Hz will not result in
> network latencies dropping by 9ms.  I know because I tried it on a server.
>

That's what I was concerned of.

> The only case where network latency improved with a higher HZ was with a
> game server (Counter-Strike) and that's only because that game's server
> component is timer dependent when calculating the game world (to do 1000
> updates per second it needs a HZ value of 1000).  So that application has
> some real-time demands, meaning a high HZ value helps.  Otherwise, you can
> stick to 100Hz on a server.  Higher values won't change anything and will
> only reduce throughput (though not by much anyway, which is why some people
> set 1000Hz even on servers.)
>

Hmmm... I *do* feel better response (interaction-wise via SSH) if I
use >100, so I think I'll settle for 250 Hz.

Thanks for the explanation!

Rgds,
-- 
FdS Pandu E Poluan
~ IT Optimizer ~

 • Blog : http://pepoluan.tumblr.com
 • Linked-In : http://id.linkedin.com/in/pepoluan

Reply via email to