Hello!
I just read through the whole undervolting thread and I've successfully done it with results of .925 at 4x and 1.075 at 8x. I tried going one step down with no immediate errors, but have to test it longer to make sure it's stable. My question is, with RM Clock 2.15, when I'm on battery power, the processor seems to stay at 1600. I've tried setting the profile to "Use P-State Transitions" but I have to select one of the FIDs and it only lets me select one. If I leave it without the transitions, will it ever throttle down to 800 mhz? I left it sitting there on battery power with nothing running and it stayed at 1600. I know I can force it to stay at 800, but I wanted the option of it bumping to 1600 if it needed it. Thanks for any help you can offer! And by the way... great undervolting thread!!
-
hi. as im new in this undervolting thingy,whats the main point of undervolting the processor? battery life? will it cause the system to become unstable to do so?
-
Hi guys,
Undervoltiong - lowering the voltage CPU is working on. Intel Pentium M for example has a Speedstep technology that changes the multiplier on FSB (front side bus) and thus changes the frequency of CPU. (Check my article on overclocking in my sig). Meaning if the bus is running at 133MHz and x14 multiplier will make CPU go at 1,86GHz. If CPU has nothing to do (idle) - it will throttle down to 133MHz x6 = 800MHz. Now at 1,86GHz Pentium M 750 will need around 1,35V to run stable on default. At 800 Mhz it will use only 0,9V or so. The higher the voltage, the higher is stability, but heat production too. So there are programs like RMClock and Notebook Hardware Control that can change the voltage per every step (multiplier) of the CPU speed, making it actually run at lower voltages at all times. 0,2V lower on every step is something that most peole do. Heat, and temperature will drop (sometimes even like 10 degrees in games), but you may encounter instability too. Once you successfully test all voltages and there are no BSODs, freezes or anything - you successfully undervolted your CPU. The speed isn't affected - only the temperature goes down and battery time could go up.
And to jabber - I don't use RM clock but in NHC - it must be set to dynamic switching (speedstep on Pentiums). Other thing NHC has is like "use only lowest and highest multiplier" setting - meaning no need for setting other transitional steps as CPU uses lowest and the highest multiplier 99,9% of the time anyway. Maybe that helps.
Cheers,
Ivan -
Jabber, for a single core CPU (Turion, I guess), RMClock 1.8 is great. I found out that higher versions of RMClock (I haven't tried 2.15, I think) introduce instability at stable voltages with 1.8ver.!?
Anyway, as Ivan said, look for an option that sounds like "dynamic switch", "automatic transition", etc. "Use P-state transitions" simply enables the throttling function of RMClock but does no define how to use it. If you turn it off, you are effectively turining off RMClock. -
-
-
Jabber,
There are two tags in each power scheme. One is call AC, the other is call Battery. You probably already setup AC scheme but forgot to do the same thing in battery tag.
Actually, you don't need to copy everything to battery tag. My battery only have lower half of my AC setup. My AC FIDs are from 4x ~ 10x, but my battery FIDs only have 4x~6 or 8x. I don't have to put full speed when I am on battery. 4x is enough for me to play DVD. -
Thanks for all the replies everyone! I was trying to use the power saving profile rather than performance on demand. That was where my problem was. I switched to Performance on Demand and now it seems to be working. I'm running on battery now and it is running at 800 Mhz! So far, I haven't had any stability issues with 2.15. I will keep trying it and see. If 1.8 is better, does anyone know where I can find it? Thanks again for the help guys! This site is great! And to ewanlin79, here's the link to the undervolting thread!
http://forum.notebookreview.com/showthread.php?t=20249
Undervolting with RM Clock 2.15
Discussion in 'HP' started by jabber, Oct 25, 2006.