Underclocking your Core Duos (max, not min V)

I realize there's a bottom end of 0.950V for CDs but I'm concerned with the higher speeds.

I've already decreased the voltage on my A8Jm 14% from 1.250 to 1.075V for its full 1.83GHz operation and it still dual primes stable. I've yet to see instability. At this voltage the fan usually stays on low and occasionally spins up to high.

The fact that I'm running full speed and my voltage is halfway between the min and max speed voltages suggests that the bottom end could go alot lower than 0.950V.

That, in turn, would appear to suggest that there must be some reason why Intel would place a ceiling at 0.950V. Perhaps so that their LV models can actually sell? It seems odd is all.

How are you guys doing? Have you tried at all?

 

The stupid thing won't fail Prime95, so I'm going to test 1.83GHz operation at 0.950V.

If this doesn't crash then it would seem the only logical conclusion is that it isn't running at 1.83GHz even though RMClock clearly says 1.83, 0.950V. The fact that I could type this message is indicative of my not having crashed yet. And btw, low fan speed, 47/53 deg. Celsius.

EDIT: One instance of P95 crashed at 0.950V, so I'll try at 1.000
EDIT2: Dual Prime95, each with in place FFTs with size=2048, Real-Time priority, has been going stable for like an hour at 1V, 1.83GHz. Temperature is 47/52C with fan consistently at low. I'll probably leave it that way overnight and see if it crashes.

 

I've had my 1.66GHz T2300 running at Vmax 1.05V for the past 6 months. This gives a useful reduction in heat generated when the CPU is fully loaded. I'm also sure that the CPU would safely run at minimum speed at significantly lower than 0.95V which would give a useful extension to operating time on battery (my previous Dothan CPU ran happily at 0.7V at minimum speed). I share your suspicion that Intel has locked the minimum voltage in order to protect its low voltage CPU sales.

John

 

Are you running 1 or 2 instances of Prime95?

 

2, both on the "max FPU stress" test, with each assigned to a different CPU. 1.000V has been going for a while now, low fan speed consistently and 47/52C.

I'm actually somewhat irritated that if I'd been able to go below 0.95V I could probably reach 3h of battery life...right now I'm stuck at about 2:40-45.

Either way, like John said at least I've got a really cool and quiet laptop now.

 

how do you Under Voltage on the A8jm???

Is this achieved from the Bios??

 

That's very good, the best I can do stable is 1.075v with my T2400, at 1.050v the laptop shuts down while Priming. I haven't played with min voltage yet but I'll let you know if I can play with it.

BTW, you may want to use the custom setting in Prime95 because max FPU stress test only uses 1024K and you may know, the T2400 has a 2MB cache, so use the custom setting and put 2048K in the max FFT and don't forget to check run FFTs in place, so you only test the CPU. And if you really want to make sure it's stable, make sure you change the Priority setting in the program to 10 to really torture the CPU,

I use CrystalCPUID to undervolt cause it gives me the option of booting with that voltage but it displays the wrong voltage. 1.404v=1.250v, so my 1.075v is actually 1.164v in that program but you can CPU-Z to see the actually voltage though.

 

Do you really see a battery life increase clocking a duo down that much? Are you talking about setting a performance cap so that no matter how much load you put on the CPU you won't throttle back up, and so save battery power?

 

My P-M 2.0Ghz runs 2.0 @ 1.100, 800Mhz @ .700.

 

Just some general information for those who may not know what Undervolting is.

To undervolt the CPU means to have it take in less power from the battery, but keep the same performance.
Yes this mean you give the CPU less power, which is measured in volts, and it will still operate at full speed, or what throttle speed you want.

What happens after undervolting? Here are some generalizations from undervolting; a cooler running laptop, less heat from particular hot spots, less fan usage and noise and longer batterylife. (Some experiencing and increase of 30 minutes)

The CPU that made this very possible was the Dothan, yes the Pentium-M, for those who don't know the very first Centrino. Many users figured out that they were able to undervolt their CPUs to achieve the above mentioned characteristics, making their laptop better for the user.

The next CPU was the Yonah, Core Duo, unfortunately they had locked voltages when you throttle the cpu to a lower Clockspeed. However this does not stop people from undervolting at higher clockspeeds.

The latest CPU is the Merom, Core 2 Duo, i have not heard much about this and its undervolting potential, like many we are interested to know

Lastly, there are 2 very popular programs people use, RM Clock and NHC, use what ever is best for you and good luck.

 

Keep in mind that on battery you're usually at minimum speed and minimum voltage, which u can't decrease on a C-D.

I would guess that you would see an increase if you use higher frequencies on battery, but I keep it at minimum voltage so that's not the case. Though, if max voltage is only 0.05V more then I think I'll switch to dynamic switching. That way, everything gets done faster and the CPU can fall into a deeper sleep state.

And no, there is no performance cap. I'm running at full performance, just with less power.

 

I have two T2400s. One is double prime stable at 1.00v. The other is double prime stable at .950v. Each chip will have a different tollerance for undervolting.

But to the original poster, yes, it is entirely possible for a a good T2400 to be stable at full load at .950v

 

I could get stable voltage at 1.013 using RMClock ... 6 hours torture test , 2 instances of Prime95 ( one on each core ).... at 1.000 it crashed after 2 hours... so, right now my max is 1.025 just to be safe... (T2400, 1.83).

Not much battery life increase at all... which is expected because the CPU is mostly running at the lowest / lower steppings most of the time.

 

Just a slight correction here: Very first Centrino was Banias core, followed by Dothan core.

 

as for battery life it is best to test abit more with the lower multipliers like 6,7,8X becauese you will use thous the most when on battery .and if you can set them as low as possible you will gain the most of battery

 

The guy who coded RMClock wrote, in a post at the RMClock forum, that it was a fairly arbitrary decision by Intel to require that voltage could not be reduced below the boot voltage. I think he had lobbied a German Intel rep to have the restriction removed for Merom, but I am not sure whether that eventuated.

I always run with the lowest voltage possible; not so much for the battery life increase (I doubt that it is even measurable) but because a 30% reduction in voltage makes a huge difference in temperature and, correspondingly, comfort and fan noise.

Nevertheless, if I had the option to either keep clock speeds fixed and reduce voltage or keep voltage fixed and increase clock speeds, I would always choose the latter