Undervolting your Toughbook?

OK,
in light of the "vote for us" thread, I looked at the other interesting threads up for consideration.

Has anyone tried undervolting on a toughbook?
any real advantage other than battery life?
disadvantages?

I did an advanced search and saw that toyo and their CF-30 had popped up in the thread some where, but with no real info on success.

anyone? bueller? bueller?

 

Ummmm, Dwagon is the tech...info falls off him like scales. BUT in my neck of the woods undervolting is heat, .....charging is proportional by that I mean the lower the voltage of a bank of batteries the longer x2 or x4 x16 it takes to bring them to their peak. (I read this)
As we speak my wind charger is keeping my batteries warm. 25 MPH not quite keeping up.
I guess the difference is I don't have Ready Kilowatt at my beck and call. If I need power...I either wait for daylight or put on my boots and parka and go start the generator. If you can without effort tie into a energy source then "by all means" play with under volts and overclocks.
What was the question? O.K. my answer is use the equipment the way it was designed. Somebody thought about it before they built it.

 

I think interestingfellow means undervolting of CPU only )

 

I have successfully undervolted my CF-29 mk3, I run the cpu (a pentium m 738)
using 0.7V for 600MHz instead of 0.988V
...
up to 0.908V for 1.4GHz instead of 1.116V

0.7V is the minimal voltage for all pentium m cpus.
I never tried to use even lower voltage for 1.4GHz

 

Ever heard of that saying "bigger is better"?

More voltage is better...

I'm sure Nikolos Tesla would agree

 

Not in this case , higher voltages is no required if stability is good al low voltage. In fact, higher voltage will cause unnecessary overheating and increase current drain from the battery, and this little voltage decrement of about 0.2-0.3 volts, at the current that the processor use, mean some watts.

 

My apologies if my answer appeared smart-alecky. I read the above mentioned post. It appears I am doing a similar thing every day. Though I am not under-volting as such I do use a Linux program to reduce CPU performance. Four settings -- conservative,on-demand,performance and power save. I have my system set up on-demand and it regularly runs at 800MHz with spikes to 2.27GHz as needed.

bueller...I was messing with I.F. again

 

It's all good.

Looks like there is no outlandish "NO!" answers around. I think I'll try it.
Benefits would include cooler cpu, longer lasting cpu (as in, until it totally craps out), and longer lasting charge.

 

that's the cpu governer and the policy 'ondemand'.
It uses the dynamic frequency scaling features of the processor i.e. (Enhanced) Intel Speedstep / ATI PowerNow - to automatically adjust the cpu freq. (power states) and voltage, based on load. On dual and multi-core modern cpus, it's independent for each core.
It's the same thing with GPUs.

Undervolting means lowering the voltage on (each) power states (at the coresponding frequency), in order to reduce heat, noise (fan), power usage while increasing battery life and ... comfort (try to use a Lenovo X100e at full load on your lap).
It's a trial and error process to find the lowest stable voltage, and time consuming, using stress testing programs at each step.

I just played a bit, trying to undervolt my previous laptop (x120e), but didn't spend enough time to get past the 'frozen system' moments.

@vadaszi: impressing tweaking you got there, what are the figures before and after (temp, wattage, battery) ?

 

Undervolting doesn't have a very large impact on idle power, e.g. at 600MHz my cpu only runs ca. 2°C cooler at idle by undervolting from 0.988 to 0.7V, but under load I can see about 10°C difference in cpu temp.

At 1.4GHz idle power consumption is only slightly higher than at 600MHz, but under load the undervolting helps a lot (before undervolting temps would quickly climb above 60°C, now temps rise much slower).

I never measured battery life under load so no numbers there (as said before the effect of undervolting on idle is negligible (for the pentium m), the difference between linux versions is much larger)

No 'frozen system' for me yet on my CF-29
For reference the numbers I used (without extensive stress testing, I simply tried them and used each frequency setting for some time):

Fid: Frequency as in: 100Mhz * Fid
Vid: Voltage as in: 0.7V + (0.016V * Vid)
FID VID
14 13
13 12
12 11
11 10
10 6
9 5
8 4
6 0

They are slightly above typical values found for the pentium m 738 on the web.
I haven't tried to reach lower values yet (since I use 600MHz most of the time).

Most noticable effect for me: even after compiling for 3h the case only got warm on the right side @600Mhz.

One of the best articles on undervolting the pentium m;
Gentoo Wiki Archives - Pentium_M/Undervolting

 

@ Vadaszi

I'm running ubuntu 11.4 on a cf28 machine, do you know if it can be slowered dinamically ? I mean with a cpu scaling applet ?

And do you know how i can undervoltage it ? I think you havn't used bios, right ?

 

@dukeluca86
Linux has a cpufreq driver for the speedstep feature in pentium III m cpus, the usual applets for frquency scaling should be usable (I'd expect it to be loaded automagically on ubuntu). A good monitoring tool for linux especially when using intel cpus/chipsets is the command line tool powertop.
It seems like the Pentium III mobile cannot be undervolted in software only when there is bios supported or a dip switch for the voltage on the mainboard.

The Pentium m (e.g. in the CF-29) can be controlled in software by writing a MSR (modell specific register) entry where the first byte is the frequency (FID) and the second byte is the voltage (VID).