Adventures in R61 Undervolting

Just thought I would share some of my undervolting experiences in case any of you guys were curious or maybe wanting to try something similar yourself. I am often on the go with my laptop so any bit of power saving that can improve my battery life is always a welcome addition in my mind. Plus, a well-undervolted processor has longer life since heat, the enemy of electronics, is reduced.

First of all, the programs I used:
RightMark Clock (RM Clock) 2.35 -- changes multipliers and voltages
Orthos2004 MultiThreaded Version -- stress test
CPU-Z 1.44.2 -- monitor voltage and multiplier separate from RM Clock

Introductory Information:
My processor is a T5450 that runs at 1.66GHz at the highest multiplier and has 2MB of L2 cache. The available multipliers on this processor range from 6X to 10X in increments of 1. I set fan speed in the power settings to maximum fan performance for this experiment since I would be stress testing my settings for long periods of time. If you're curious about how RM Clock functions, flipfire wrote this handy article detailing its uses here

To start with, it seems that Lenovo/Intel set my voltages at ~1.0V for 6X multiplier (1.0GHz operating speed), ~1.15V for 8X multiplier (1.33GHz operating speed) and ~1.25V for 10X multiplier (1.66GHz operating speed). This resulted in initial temperature readings of about 67 degrees for 1.66GHz after being loaded on Orthos Small FFT test for 30 minutes.

Since I usually am on the go with my laptop, I figured that I would start with the lower multipliers. Originally, the 8X multiplier showed 1.15V. I began by lowering that in increments of 0.05V and testing on Orthos Small FFT for 1 hour each time. Eventually, I bottomed out at 0.95V. I then moved on to the 10X multiplier since I was only planning on using 3 power states, 6X, 8X, and 10X. The 10X started out at 1.25V. Lowering by 0.05V and testing on Orthos again netted me 0.95V!

Overall, I've been pretty satisfied with these results although I wish there were options available for me to take this fantastic processor even lower. If the 10X can do 0.95V, the 6X must be able to do at least 0.6V! I could have my own LV processor for free...but alas, there are no lower voltages allowed by the BIOS. At these voltages, I have seen the temperature go from 67 degrees to 60 degrees at full load. Battery life has improved by 30 minutes.


Quick Recap:
Starting:
6X @ 0.95V
8X @ 1.15V
10X @ 1.25V 67 degrees Celsius

Final:
6X @ 0.95V
8X @ 0.95V
10X @ 0.95V 60 degrees Celsius

 

Thanks for sharing! You got a good CPU there. The voltages set by Intel is to ensure 100 % stability in all possible aspects for all manufactured CPU's of a certain model.

 

In a similar adventure, I have my T7200 for .95V at 6x, and then .975 at the highest multiplier. Ill try .95 across the board, maybe that might squeeze another 10-15 mins! I also got about a half hour improvement going from the unmanaged voltages to the .950->.975 range.

4.5 hrs battery life baby!

 

Is it safe to assume that if at the highest multiplier, you are able to get 0.975V, then it's safe to set all the multiplier levels to 0.975V?

 

All this is very intriguing but is there a way to set the voltages in the BIOS?

 

Yes but its still a good idea to do a stability test for all the multipliers except for 6x (idle)

No, only desktops can.