On 24/12/2010, at 1:54 PM, Jam wrote:

> On Friday 24 December 2010 09:00:03 [email protected] wrote:
>>> PS - once jester is doing basic snmp/cacti/squid/etc and the QNAP doing
>>> the "heavy-lifting", I intend under-clocking the little dear from
>>> 2.66GHz to about 1.6GHz :)  Save the planet and all that. I wouldn't
>>> under clock it, it should support dynamic clocking (one presumes) that
>>> will drop the clock pretty low when idle, then bump it up when in use.
>>> This often will save power as the CPU can stay in low power "sleep"
>>> states for longer, at least on the smaller end of the scale.
>> 
>> Yep - will do that as well as under-clocking (which is the plan).  THe
>> under-clocking will also reduce heat and power on the FSB etc. too, which
>> wouldn't happen with dynamic scaling alone AFAIK.  Happy to be proven
>> wrong on this as the CPU barely breaks out of "brrr, I'm freezing, can
>> someone please do something to warm me up" mode, and rarely heats up much
>> beyond ambient, so if I can avoid the under-clocking....GREAT!  The two
>> 15k RPM 3.5" 500GB drives though, they generate a metric truck load of
>> heat!!  Hence the reason I'm ditching them for an SSD and external NAS.
> 
> Actually the thought experiment is krap.
> Perth: outside 33C inside 32C cpu clocked normally and doing 'normal' stuff 
> (a 
> couple of VMs, some mythbackend, a firefox or two, kmail, gnome desktop) is 
> idling along at 1G and 37C
> 
> [haycorn] /home/jam [1999]% cat /proc/cpuinfo
> processor     : 0
> vendor_id     : AuthenticAMD
> cpu family    : 15
> model         : 107
> model name    : AMD Athlon(tm) X2 Dual Core Processor BE-2350

Cool :)  Here's the results from my fat server after some mild load doing some 
big DB queries for the last 5min:
processor       : 0-3 (edited by moi)
vendor_id       : AuthenticAMD
cpu family      : 16
model           : 2
model name      : AMD Phenom(tm) 9650 Quad-Core Processor

Ambient now is 26C.

# sensors
atk0110-acpi-0
Adapter: ACPI interface
Vcore Voltage:   +1.12 V  (min =  +0.85 V, max =  +1.60 V)
 +3.3 Voltage:   +3.34 V  (min =  +2.97 V, max =  +3.63 V)
 +5 Voltage:     +5.07 V  (min =  +4.50 V, max =  +5.50 V)
 +12 Voltage:   +12.16 V  (min = +10.20 V, max = +13.80 V)
CPU FAN Speed:  2376 RPM  (min =  600 RPM)
CPU Temperature: +43.0°C  (high = +60.0°C, crit = +95.0°C)  
MB Temperature:  +41.0°C  (high = +45.0°C, crit = +95.0°C)

# for drive in sda sdb
> do
> smartctl -A /dev/$drive | grep 194
> done
194 Temperature_Celsius     0x0022   104   084   000    Old_age   Always       
-       43
194 Temperature_Celsius     0x0022   107   087   000    Old_age   Always       
-       40

Mind you, I have a trio of 25mm fans screaming at 6000RPM just to keep the two 
hard drives below 60C!  So even with seriously annoying cooling (another reason 
to ditch the 15K drives)  and relatively cool day, the CPU ran up from 
29C->43C, but the HDDs barely change with extra I/O.  If I throw something 
bigger at it, like transcoding streaming media to my PS3, the CPU will easily 
hit 70C...hence the reason I don't do it much.

I'm interested in seeing how much (if any) power consumption drops (and 
corresponding heat) with under-clocking on consumer-grade kit.  I haven't seen 
an quantitative studies on the subject so I want to do the experiment myself :) 
 Every pushes for bigger/faster/more...I'm curious about taking big/fast/more 
and putting the skids on it.

Thanks for the input.

Cheers,

James

Attachment: smime.p7s
Description: S/MIME cryptographic signature

-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Reply via email to